back to indexNick Lane: Origin of Life, Evolution, Aliens, Biology, and Consciousness | Lex Fridman Podcast #318
link |
Well, the source of energy at the origin of life
link |
is the reaction between carbon dioxide and hydrogen.
link |
And amazingly, most of these reactions are exergonic,
link |
which is to say they release energy.
link |
If you have hydrogen and CO2,
link |
and you put them together in a falcon tube
link |
and you warm it up to say 50 degrees centigrade,
link |
and you put in a couple of catalysts and you shake it,
link |
nothing's gonna happen.
link |
But thermodynamically, that is less stable.
link |
Two gases, hydrogen and CO2, is less stable than cells.
link |
What should happen is you get cells coming out.
link |
Why doesn't that happen?
link |
It's because of the kinetic barriers.
link |
That's where you need the spark.
link |
The following is a conversation with Nick Lane,
link |
a biochemist at University College London
link |
and author of some of my favorite books
link |
on biology, science, and life ever written,
link |
including his two most recent titled
link |
"'Transformer,' The Deep Chemistry of Life and Death,"
link |
and the vital question, why is life the way it is?
link |
This is the Lex Friedman podcast.
link |
To support it, please check out our sponsors
link |
in the description.
link |
And now, dear friends, here's Nick Lane.
link |
Let's start with perhaps the most mysterious,
link |
the most interesting question that we little humans
link |
can ask of ourselves.
link |
How did life originate on Earth?
link |
You could ask anybody working on the subject,
link |
and you'll get a different answer from all of them.
link |
They will be pretty passionately held opinions,
link |
and their opinions grounded in science,
link |
but they're still really, at this point, their opinions,
link |
because there's so much stuff to know
link |
that all we can ever do is get a small slice of it,
link |
and it's the context which matters.
link |
So I can give you my answer.
link |
My answer is from a biologist's point of view.
link |
That has been missing from the equation over decades,
link |
which is, well, what does life do on Earth?
link |
Why is it this way?
link |
Why is it made of cells?
link |
Why is it made of carbon?
link |
Why is it powered by electrical charges on membranes?
link |
There's all these interesting questions about cells
link |
that if you then look to see,
link |
well, is there an environment on Earth,
link |
on the early Earth four billion years ago,
link |
that kind of matches the requirements of cells?
link |
Well, there is one.
link |
There's a very obvious one.
link |
It's basically created by whenever you have
link |
a wet, rocky planet, you get these hydrothermal vents,
link |
which generate hydrogen gas in bucket loads,
link |
and electrical charges on kind of cell like pores
link |
that can drive the kind of chemistry that life does.
link |
So it seems so beautiful and so obvious
link |
that I've spent the last 10 years or more
link |
trying to do experiments.
link |
It turns out to be difficult, of course.
link |
Everything's more difficult than you ever thought
link |
it was gonna be, but it looks, I would say,
link |
more true rather than less true over that 10 year period.
link |
I think I have to take a step back every now and then
link |
and think, hang on a minute, where's this going?
link |
I'm happy it's going in a sensible direction.
link |
And I think then you have these other interesting dilemmas.
link |
I mean, I'm often accused of being too focused
link |
on life on Earth, too kind of narrow minded
link |
and inward looking, you might say.
link |
I'm talking about carbon, I'm talking about cells,
link |
and maybe you or plenty of people can say to me,
link |
ah, yeah, but life can be anything.
link |
I have no imagination.
link |
And maybe they're right.
link |
But unless we can say why life here is this way,
link |
and if those reasons are fundamental reasons,
link |
or if they're just trivial reasons,
link |
then we can't answer that question.
link |
So I think they're fundamental reasons,
link |
and I think we need to worry about them.
link |
Yeah, there might be some deep truth to the puzzle
link |
here on Earth that will resonate
link |
with other puzzles elsewhere that will,
link |
solving this particular puzzle
link |
will give us that deeper truth.
link |
So what, to this puzzle, you said vents, hydrogen,
link |
wet, so chemically, what is the potion here?
link |
How important is oxygen?
link |
You wrote a book about this.
link |
Yeah, and I actually just came straight here
link |
from a conference where I was chairing a session
link |
on whether oxygen matters or not in the history of life.
link |
Of course it matters, but it matters most
link |
to the origin of life to be not there.
link |
As I see it, we have this, I mean,
link |
life is made of carbon, basically, primarily,
link |
organic molecules with carbon, carbon bonds.
link |
And the building block, the Lego brick
link |
that we take out of the air or take out of the oceans
link |
is carbon dioxide.
link |
And to turn carbon dioxide into organic molecules,
link |
we need to strap on hydrogen.
link |
And so we need, and this is basically
link |
what life is doing, it's hydrogenating carbon dioxide.
link |
It's taking the hydrogen, the bubbles out of the earth
link |
in these hydrothermal vents, and it sticks it on CO2.
link |
And it's kind of really as simple as that.
link |
And actually, thermodynamically,
link |
there's the thing that I find most troubling
link |
is that if you do these experiments in the lab,
link |
the molecules you get are exactly the molecules
link |
that we see at the heart of biochemistry
link |
in the heart of life.
link |
Is there something to be said about the earliest origins
link |
of that little potion, that chemical process?
link |
What really is the spark there?
link |
There isn't a spark.
link |
There is a continuous chemical reaction.
link |
And there is kind of a spark,
link |
but it's a continuous electrical charge
link |
which helps drive that reaction.
link |
There's a literally spark.
link |
Well, the charge at least, but yes.
link |
I mean, a spark in that sense is,
link |
we tend to think of in terms of Frankenstein,
link |
we tend to think in terms of electricity
link |
and one moment you zap something and it comes alive.
link |
And what does that really mean?
link |
It's come alive and now what's sustaining it?
link |
Well, we are sustained by oxygen,
link |
by this continuous chemical reaction.
link |
And if you put a plastic bag on your head,
link |
then you've got a minute or something before it's all over.
link |
So some way of being able to leverage a source of energy.
link |
Well, the source of energy at the origin of life
link |
is the reaction between carbon dioxide and hydrogen.
link |
And amazingly, most of these reactions are exergonic,
link |
which is to say they release energy.
link |
If you have hydrogen and CO2
link |
and you put them together in a falcon tube
link |
and you warm it up to say 50 degrees centigrade
link |
and you put in a couple of catalysts and you shake it,
link |
nothing's gonna happen.
link |
But thermodynamically, that is less stable.
link |
Two gases, hydrogen and CO2, is less stable than cells.
link |
What should happen is you get cells coming out.
link |
So why doesn't that happen?
link |
It's because of the kinetic barriers.
link |
That's where you need the spark.
link |
Is it possible that life originated
link |
multiple times on Earth?
link |
The way you describe it, you make it sound so easy.
link |
There's a long distance to go from the first bits
link |
of prebiotic chemistry to, say, molecular machines
link |
Is that the first thing that you would say is life?
link |
Like if I introduce you, the two of you at a party,
link |
you would say that's a living thing?
link |
I would say as soon as we introduce genes, information,
link |
into systems that are growing anyway,
link |
so I would talk about growing protocells,
link |
as soon as we introduce even random bits of information
link |
into there, I'm thinking about RNA molecules, for example,
link |
doesn't have to have any information in it.
link |
It can be a completely random sequence.
link |
But if it's introduced into a system which is in any case
link |
growing and doubling itself and reproducing itself,
link |
then any changes in that sequence that allow it
link |
to do so better or worse are now selected
link |
by perfectly normal natural selection.
link |
But it's a system.
link |
So that's when it becomes alive to my mind.
link |
That's encompassed into like an object
link |
that keeps information and evolves that information
link |
over time or changes that information over time
link |
in response to the.
link |
So it's always part of a cell system
link |
from the very beginning.
link |
So is your sense that it started only once
link |
because it's difficult or is it possibly started
link |
in multiple locations on Earth?
link |
It's possible it started multiple occasions.
link |
There's two provisos to that.
link |
One of them is oxygen makes it impossible really
link |
for life to start.
link |
So as soon as we've got oxygen in the atmosphere,
link |
then life isn't gonna keep starting over.
link |
So I often get asked by people,
link |
why can't we have life starting?
link |
If it's so easy, why can't life start in these vents now?
link |
And the answer is if you want hydrogen to react with CO2
link |
and there's oxygen there, hydrogen reacts
link |
with oxygen instead.
link |
It's just, you get an explosive reaction that way.
link |
So it's never gonna happen.
link |
But for the origin of life earlier than that,
link |
all we know is that there's a single common ancestor
link |
There could have been multiple origins
link |
and they all just disappeared.
link |
But there's a very interesting deep split in life
link |
between bacteria and what are called archaea,
link |
which look just the same as bacteria.
link |
And they're not quite as diverse, but nearly.
link |
And they are very different in their biochemistry.
link |
And so any explanation for the origin of life
link |
has to account as well for why they're so different
link |
and yet so similar.
link |
And that makes me think that life probably
link |
did arise only once.
link |
Can you describe the difference that's interesting there?
link |
Well, how they're similar, how they're different?
link |
Well, they're different in their membranes primarily.
link |
They're different in things like DNA replication.
link |
They use completely different enzymes
link |
and the genes behind it for replicating DNA.
link |
So they both have membranes, both have DNA replication.
link |
The process of that is different.
link |
They both have DNA.
link |
The genetic code is identical in them both.
link |
The way in which it's transcribed into RNA,
link |
into the copy of a gene,
link |
and the way that that's then translated into a protein,
link |
that's all basically the same in both of these groups.
link |
So they clearly share a common ancestor.
link |
It's just that they're different
link |
in fundamental ways as well.
link |
And if you think about, well,
link |
what kind of processes could drive that divergence
link |
I can think about it in terms of membranes,
link |
in terms of the electrical charges on membranes.
link |
And it's that that makes me think that there was probably,
link |
there were probably many unsuccessful attempts
link |
but only one really successful attempt.
link |
Can you explain why that divergence
link |
makes you think there's one common ancestor?
link |
Okay, can you describe that intuition?
link |
I'm a little bit unclear about why the divert,
link |
like the leap from the divergence means there's one.
link |
Do you mean like the divergence indicates
link |
that there was a big invention at that time from one source?
link |
If you'd got, as I imagine it,
link |
you have a common ancestor living in a hydrothermal vent.
link |
Let's say there are millions of vents
link |
and millions of potential common ancestors
link |
living in all of those vents,
link |
but only one of them makes it out first.
link |
Then you could imagine that that cell
link |
is then gonna kind of take over the world
link |
and wipe out everything else.
link |
And so what you would see would be
link |
a single common ancestor for all of life.
link |
But with lots of different vent systems
link |
all kind of vying to create the first life forms,
link |
So this thing is a cell, a single cell organism.
link |
We're always talking about populations of cells, but yes.
link |
These are single celled organisms.
link |
But the fundamental life form is a single cell, right?
link |
So like, or, so they're always together
link |
but they're alone together.
link |
There's a machinery in each one individual component
link |
that if left by itself would still work, right?
link |
It's the unit of selection is a single cell.
link |
But selection operates over generations
link |
and changes over generations in populations of cells.
link |
So it would be impossible to say that a cell
link |
is the unit of selection in the sense that
link |
unless you have a population, you can't evolve,
link |
Right, but there was one Chuck Norris,
link |
it's an American reference cell
link |
that made it out of the vents, right?
link |
Or like the first one.
link |
So imagine then that there's one cell gets out
link |
and it takes over the world.
link |
It gets out in the water, it's like floating around.
link |
We're deep in the ocean somewhere.
link |
But actually two cells got out
link |
and they appear to have got out from the same vent
link |
because they both share the same code and everything else.
link |
So unless all the, you know,
link |
we've got a million different common ancestors
link |
in all these different vents.
link |
So either they all have the same code
link |
and two cells spontaneously merge from different places
link |
or two different cells, fundamentally different cells
link |
came from the same place.
link |
So either way, what are the constraints that say,
link |
not just one came out or not half a million came out,
link |
That's kind of a bit strange.
link |
So how did they come out?
link |
Well, they come out because what are you doing inside a vent
link |
is you're relying on the electrical charges down there
link |
to power this reaction between hydrogen and CO2
link |
to make yourself grow.
link |
And when you leave the vent, you've got to do that yourself.
link |
You've got to power up your own membrane.
link |
And so the question is,
link |
well, how do you power up your own membrane?
link |
And the answer is, well, you need to pump.
link |
You need to pump ions to give an electrical charge
link |
So what do the pumps look like?
link |
Well, the pumps look different in these two groups.
link |
It's as if they both emerged from a common ancestor.
link |
As soon as you've got that ancestor,
link |
things move very quickly and divergently.
link |
Why does the DNA replication look different?
link |
Well, it's joined to the membrane.
link |
The membranes are different.
link |
The DNA replication is different
link |
because it's joined to a different kind of membrane.
link |
So there's interesting, this is detail, you may say,
link |
but it's also fundamental
link |
because it's about the two big divergent groups
link |
of life on Earth that seem to have diverged really early on.
link |
And it all started from one organism.
link |
And then that organism just start replicating
link |
the heck out of itself with some mutation of the DNA.
link |
So like there's some, there's a competition
link |
through the process of evolution.
link |
They're not like trying to beat each other up.
link |
They're just, they're just trying to live.
link |
Just replicate us.
link |
Well, you know, let's not minimize there.
link |
They're just trying to chill.
link |
They're trying to relax up.
link |
There's no, but there's no sense of trying to survive.
link |
They're replicating.
link |
I mean, there's no sense
link |
in which they're trying to do anything.
link |
They're just kind of an outgrowth of the Earth,
link |
Of course, the aliens would describe us humans
link |
They might be right.
link |
This primitive life.
link |
It's just ants that are hairless,
link |
What do you think about the idea of panspermia
link |
that the theory that life did not originate on Earth
link |
and was planted here from outer space?
link |
Or pseudopanspermia, which is like the basic ingredients,
link |
the magic that you mentioned was planted here
link |
from elsewhere in space?
link |
I don't find them helpful.
link |
That's not to say they're wrong.
link |
So, pseudotranspermia, the idea that the chemicals,
link |
the amino acids, the nucleotides
link |
are being delivered from space.
link |
Well, we know that happens.
link |
They're delivered on meteorites, comets, and so on.
link |
So what do they do next?
link |
That's, to me, the question.
link |
Well, what do they do is they stock a soup.
link |
Presumably, they land in a pond or in an ocean
link |
or wherever they land.
link |
And then you end up with a best possible case scenario
link |
is you end up with a soup of nucleotides
link |
And then you have to say, so now what happens?
link |
And the answer is, oh, well, they have to go,
link |
bloop, become alive.
link |
So how did they do that?
link |
And you may as well say then a miracle happened.
link |
I don't believe in soup.
link |
I think what we have in event is a continuous conversion,
link |
a continuous growth, a continuous reaction,
link |
a continuous converting a flow of molecules
link |
into more of yourself, you might say,
link |
even if it's a small bit.
link |
So you've got a kind of continuous self organization
link |
and growth from the very beginning.
link |
You never have that in a soup.
link |
Isn't the entire universe and living organisms
link |
in the universe, isn't it just soup all the way down?
link |
Isn't it all soup?
link |
No, no, I mean, soup almost by definition
link |
doesn't have a structure.
link |
But soup is a collection of ingredients
link |
that are like randomly interacting.
link |
Yeah, but they're not random.
link |
They're not, I mean, we have chemistry going on here.
link |
We have metal grains forming, which are, you know,
link |
effective oil water interactions.
link |
Okay, so it feels like there's a direction to a process,
link |
like a directed process.
link |
There are directions to processes, yeah.
link |
And if you're starting with CO2
link |
and you've got two reactive fluids being brought together
link |
and they react, what are they gonna make?
link |
Well, they make carboxylic acids,
link |
which include the fatty acids
link |
that make up the cell membranes.
link |
And they form directly into bilayer membranes.
link |
They form like soap bubbles.
link |
It's spontaneous organization caused by the nature
link |
And those things are capable of growing
link |
and are capable in effect of being selected
link |
even before there are genes.
link |
We have this, so we have a lot of order
link |
and that order is coming from thermodynamics.
link |
And the thermodynamics is always about increasing
link |
the entropy of the universe.
link |
But if you have oil and water and they're separating,
link |
you're increasing the entropy of the universe,
link |
even though you've got some order,
link |
which is the soap and the water are not missable.
link |
Now, to come back to your first question
link |
about panspermia properly,
link |
that just pushes the question somewhere else.
link |
That just, even if it's true,
link |
maybe life did start on Earth by panspermia.
link |
So what are the principles
link |
that govern the emergence of life on any planet?
link |
It's an assumption that life started here.
link |
And it's an assumption that it started
link |
in a hydrothermal vent or it started
link |
in a terrestrial geothermal system.
link |
The question is, can we work out a testable sequence
link |
of events that would lead from one to the other one
link |
and then test it and see if there's any truth in it or not?
link |
With panspermia, you can't do any of that.
link |
But the fundamental question of panspermia is,
link |
do we have the machine here on Earth to build life?
link |
Is the vents enough?
link |
Is oxygen and hydrogen and whatever the heck else we want
link |
and some source of energy and heat,
link |
is that enough to build life?
link |
Well, that's, of course you would say that as a human.
link |
But there could be aliens right now
link |
chuckling at that idea.
link |
Maybe you need some special sauce,
link |
special elsewhere sauce.
link |
So your sense is we have everything here.
link |
I mean, this is precisely the question.
link |
I like to, when I'm talking in schools,
link |
I like to start out with the idea
link |
of we can make a time machine.
link |
We go back four billion years
link |
and we go to these environments that people talk about.
link |
We go to a deep sea hydrothermal vent,
link |
we go to a kind of Yellowstone Park type place environment
link |
and we find some slime that looks like we can test it.
link |
It's made of organic molecules.
link |
It's got a structure which is not obviously cells,
link |
but you know, is this a stepping stone
link |
on the way to life or not?
link |
Unless we've got an intellectual framework
link |
that says this is a stepping stone and that's not a step.
link |
You know, we'd never know.
link |
We wouldn't know which environment to go to,
link |
what to look for, how to say this.
link |
So all we can ever hope for,
link |
because we're never gonna build that time machine,
link |
is to have an intellectual framework
link |
that can explain step by step, experiment by experiment,
link |
how we go from a sterile inorganic planet
link |
to living cells as we know them.
link |
And in that framework, every time you have a choice,
link |
it could be this way or it could be that way,
link |
or there's lots of possible forks down that road.
link |
Did it have to be that way?
link |
Could it have been the other way?
link |
And would that have given you life
link |
with very different properties?
link |
And so if you come up with a, you know,
link |
it's a long hypothesis, because as I say,
link |
we're going from really simple prebiotic chemistry
link |
all the way through to genes and molecular machines.
link |
That's a long, long pathway.
link |
And nobody in the field would agree on the order
link |
in which these things happened,
link |
which is not a bad thing,
link |
because it means that you have to go out
link |
and do some experiments and try and demonstrate
link |
that it's possible or not possible.
link |
It's so freaking amazing that it happened though.
link |
It feels like there's a direction to the thing.
link |
Can you try to answer from a framework perspective
link |
So you said there's some order and yet there's complexity.
link |
So it's not perfectly ordered.
link |
There's still some fun in it.
link |
And it also feels like the processes have a direction
link |
through the selection mechanism.
link |
They seem to be building something,
link |
always better, always improving.
link |
I mean, maybe it's...
link |
I mean, that's a perception.
link |
That's our romanticization of things are always better.
link |
Things are getting better, we'd like to believe that.
link |
I mean, you think about the world
link |
from the point of view of bacteria
link |
and bacteria are the first things to emerge
link |
from whatever environment they came from.
link |
And they dominated the planet very, very quickly.
link |
And they haven't really changed.
link |
Four billion years later, they look exactly the same.
link |
So about four billion years ago,
link |
bacteria started to really run the show.
link |
And then nothing happened for a while.
link |
Nothing happened for two billion years.
link |
Then after two billion years,
link |
we see another single event origin, if you like,
link |
of our own type of cell, the eukaryotic cells.
link |
So cells with a nucleus and lots of stuff going on inside.
link |
Another singular origin.
link |
It only happened once in the history of life on earth.
link |
Maybe it happened multiple times and there's no evidence.
link |
Everything just disappeared,
link |
but we have to at least take it seriously
link |
that there's something that stops bacteria
link |
from becoming more complex because they didn't.
link |
That's a fact that they emerged four billion years ago.
link |
And something happened two billion years ago,
link |
but the bacteria themselves didn't change.
link |
They remain bacterial.
link |
So there is no trajectory, necessary trajectory
link |
towards great complexity in human beings at the end of it.
link |
It's very easy to imagine that without photosynthesis
link |
arising or without eukaryotes arising,
link |
that a planet could be full of bacteria and nothing else.
link |
We'll get to that because that's a brilliant invention.
link |
And there's a few brilliant invention along the way.
link |
If you were to show up on earth,
link |
but to take that time machine,
link |
and you said, asking yourself the question,
link |
is this a stepping stone towards life?
link |
As you step along, when you see the early bacteria,
link |
how would you know it's life?
link |
And then this is really important question
link |
when you go to other planets and look for life.
link |
Like what is the framework of telling a difference
link |
between a rock and a bacteria?
link |
I mean, the question's kind of both impossible to answer
link |
and trivial at the same time.
link |
And I don't like to answer it
link |
because I don't think there is an answer.
link |
I think we're trying to describe the process of time.
link |
Those are the most fun questions.
link |
What do you mean there's no answer?
link |
No, there is no answer.
link |
I mean, there's lots of,
link |
there are at least 40 or 50 different definitions
link |
of life out there.
link |
And most of them are, well, obviously bad
link |
in one way or another.
link |
I mean, there's freaks.
link |
I can never remember the exact words that people use,
link |
but there's a NASA working definition of life,
link |
which more or less says a system,
link |
which is capable of self sustaining system,
link |
capable of evolution or something along those lines.
link |
And I immediately have a problem
link |
with the word self sustaining
link |
because it's sustained by the environment.
link |
And I know what they're getting at.
link |
I know what they're trying to say,
link |
but I pick a hole in that.
link |
And there's always wags who say,
link |
but you know, by that definition, a rabbit is not alive.
link |
Only a pair of rabbits would be alive
link |
because a single rabbit is incapable of copying itself.
link |
There's all kinds of pedantic, silly,
link |
but also important objections to any hypothesis.
link |
The real question is what is, you know,
link |
we can argue all day or people do argue all day
link |
about is a virus alive or not?
link |
And it depends on the content.
link |
Most biologists could not agree about that.
link |
So then what about a jumping gene,
link |
a retro element or something that is even simpler
link |
than a virus, but it's capable of converting
link |
its environment into a copy of itself.
link |
And that's about as close, this is not a definition,
link |
but this is a kind of a description of life
link |
is that it's able to parasitize the environment.
link |
And that goes for plants as well as animals
link |
and bacteria and viruses to make a relatively exact copy
link |
of themselves, informationally exact copy of themselves.
link |
By the way, it doesn't really have to be
link |
a copy of itself, right?
link |
It just has to be, you have to create something
link |
that's interesting, the way evolution is.
link |
So it is extremely powerful process of evolution,
link |
which is basically make a copy of yourself
link |
and sometimes mess up a little bit.
link |
That seems to work really well.
link |
I wonder if it's possible to mess up big time
link |
as a standard, as a default.
link |
It's called a hopeful monster and in principle it can.
link |
Actually, it turns out, I would say that this is due
link |
a reemergence, this is some amazing work
link |
from Michael Levin, I don't know if you came across him,
link |
but if you haven't interviewed him,
link |
you should interview him about, yeah.
link |
I'm talking to him in a few days.
link |
So I mentioned, there's two people that Andre,
link |
if I may mention, Andre Kapathe is a friend
link |
who's really admired in the AI community,
link |
said you absolutely must talk to Michael and to Nick.
link |
So of course, I'm a huge fan of yours,
link |
so I'm really fortunate that we can actually
link |
Anyway, you were saying?
link |
Well, Michael Levin is doing amazing work,
link |
basically about the way in which electrical fields
link |
control development and he's done some work
link |
with planarian worms, so flat worms,
link |
where he'll tell you all about this,
link |
so I won't say any more than the minimum,
link |
but basically you can cut their head off
link |
and they'll redevelop a different, a new head.
link |
But the head that they develop depends,
link |
if you knock out just one iron pump in a membrane,
link |
so you change the electrical circuitry just a little bit,
link |
you can come up with a completely different head.
link |
It can be a head which is similar to those
link |
that diverged 150 million years ago
link |
or it can be a head which no one's ever seen before,
link |
a different kind of head.
link |
Now that is really, you might say, a hopeful monster.
link |
This is a kind of leap into a different direction.
link |
The only question for natural selection is does it work?
link |
Is the change itself feasible as a single change?
link |
And the answer is yes, it's just a small change
link |
And the second thing is it gives rise
link |
to a completely different morphology.
link |
And if it works, that can easily be a shift.
link |
But for it to be a speciation, for it to continue,
link |
for it to give rise to a different morphology over time,
link |
then it has to be perpetuated.
link |
So that shift, that change in that one gene
link |
has to work well enough that it is selected and it goes on.
link |
And copied enough times to where you can really test it.
link |
So the likelihood, it would be lost,
link |
but there will be some occasions where it survives.
link |
And yes, the idea that we can have sudden, fairly abrupt
link |
changes in evolution, I think it's time for a rebirth.
link |
What about this idea that kind of trying to
link |
mathematize a definition of life and saying how many steps,
link |
the shortest amount of steps it takes to build the thing,
link |
almost like an engineering view of it?
link |
Ah, I like that view.
link |
Because I think that in a sense, that's not very far away
link |
from what a hypothesis needs to do
link |
to be a testable hypothesis for the origin of life.
link |
You need to spell out, here's each step,
link |
and here's the experiment to do for each step.
link |
The idea that we can do it in the lab,
link |
some people say, oh, we'll have created life
link |
within five years, but ask them what they mean by life.
link |
We have a planet four billion years ago
link |
with these vent systems across the entire surface
link |
of the planet, and we have millions of years if we wanted.
link |
I have a feeling that we're not talking about
link |
millions of years.
link |
I have a feeling we're talking about maybe millions
link |
of nanoseconds or picoseconds.
link |
We're talking about chemistry, which is happening quickly.
link |
But we still need to constrain those steps,
link |
but we've got a planet doing similar chemistry.
link |
You asked about a trajectory.
link |
The trajectory is the planetary trajectory.
link |
The planet has properties.
link |
Basically, it's got a lot of iron at the center of it.
link |
It's got a lot of electrons at the center of it.
link |
It's more oxidized on the outside,
link |
partly because of the sun and partly because the heat
link |
of volcanoes puts out oxidized gases.
link |
So the planet is a battery.
link |
It's a giant battery, and we have a flow of electrons
link |
going from inside to outside in these hydrothermal vents,
link |
and that's the same topology that a cell has.
link |
A cell is basically just a micro version of the planet,
link |
and there is a trajectory in all of that,
link |
and there's an inevitability that certain types
link |
of chemical reaction are going to be favored over others,
link |
and there's an inevitability in what happens in water,
link |
the chemistry that happens in water.
link |
Some will be immiscible with water and will form membranes
link |
and will form insoluble structures,
link |
and nobody really understands water very well,
link |
and it's another big question.
link |
For experiments on the origin of life, what do you put it in?
link |
What kind of structure do we want to induce in this water?
link |
Because the last thing it's likely to be
link |
is just kind of bulk water.
link |
How fundamental is water to life, would you say?
link |
I would say pretty fundamental.
link |
I wouldn't like to say it's impossible for life
link |
to start any other way, but water is everywhere.
link |
Water's extremely good at what it does,
link |
and carbon works in water especially well.
link |
So those things, and carbon is everywhere.
link |
So those things together make me think probabilistically,
link |
if we found a thousand life forms, 995 of them
link |
would be carbon based and living in water.
link |
Now the reverse question, if you found a puddle of water
link |
elsewhere and some carbon, no, just a puddle of water.
link |
Is a puddle of water a pretty damn good indication
link |
that life either exists here or has once existed here?
link |
So it doesn't work the other way.
link |
I think you need a living planet.
link |
You need a planet which is capable
link |
of turning over its surface.
link |
It needs to be a planet with water.
link |
It needs to be capable of bringing those electrons
link |
from inside to the outside.
link |
It needs to turn over its surface.
link |
It needs to make that water work and turn it into hydrogen.
link |
So I think you need a living planet.
link |
But once you've got the living planet,
link |
I think the rest of it is kind of thermodynamics all the way.
link |
So if you were to run Earth over a million times up
link |
to this point, maybe beyond, to the end,
link |
let's run it to the end, what is it?
link |
How much variety is there?
link |
You kind of spoke to this trajectory
link |
that the environment dictates chemically,
link |
I don't know in which other way, spiritually,
link |
I don't know, like dictates kind of the direction
link |
of this giant machine that seems chaotic,
link |
but it does seem to have order in the steps it's taking.
link |
How often will life, how often will bacteria emerge?
link |
How often will something like humans emerge?
link |
How much variety do you think there would be?
link |
I think at the level of bacteria, not much variety.
link |
I think we would get, that's how many times
link |
you say you want to run it, a million times.
link |
I would say at least a few hundred thousand will get bacteria again.
link |
Because I think there's some level of inevitability
link |
that a wet rocky planet will give rise
link |
through the same processes to something very close.
link |
I think this is not something I'd have thought
link |
a few years ago, but working with a PhD student
link |
of mine, Stuart Harrison, he's been thinking
link |
about the genetic code, and we've just been publishing
link |
on that, there are patterns that you can discern in the code,
link |
or he has discerned in the code,
link |
that if you think about them in terms of,
link |
we start with CO2 and hydrogen,
link |
and these are the first steps of biochemistry,
link |
you come up with a code which is very similar
link |
to the code that we see.
link |
So it wouldn't surprise me any longer
link |
if we found life on Mars and it had a genetic code
link |
that was not very different to the genetic code
link |
that we have here, without it just being transferred across.
link |
There's some inevitability about the whole
link |
of the beginnings of life, in my view.
link |
That's really promising, because if the basic chemistry
link |
is tightly linked to the genetic code,
link |
that means we can interact with other life
link |
if it exists out there.
link |
Well, that's potentially.
link |
That's really exciting, if that's the case.
link |
Okay, but then bacteria.
link |
We've got bacteria.
link |
How easy is photosynthesis?
link |
Much harder, I would say.
link |
Let's actually go there.
link |
Let's go through the inventions.
link |
What is photosynthesis?
link |
And why is it hard?
link |
Well, there are different forms.
link |
I mean, basically, you're taking hydrogen
link |
and you're sticking it onto CO2,
link |
and it's powered by the sun.
link |
Question is, where are you taking the hydrogen from?
link |
And in photosynthesis that we know in plants,
link |
it's coming from water.
link |
So you're using the power of the sun to split water,
link |
take out the hydrogen, stick it onto CO2,
link |
and the oxygen is a waste product,
link |
and you just throw it out, throw it away.
link |
So it's the single greatest planetary pollution event
link |
in the whole history of the Earth.
link |
The pollutant being oxygen.
link |
It also made possible animals.
link |
You can't have large, active animals
link |
without an oxygenated atmosphere,
link |
at least not in the sense that we know on Earth.
link |
So that's a really big invention
link |
in the history of Earth. Huge invention, yes.
link |
And it happened once.
link |
There's a few things that happen once on Earth,
link |
and you're always stuck with this problem.
link |
Once it happened, did it become so good so quickly
link |
that it precluded the same thing happening ever again?
link |
Or are there other reasons?
link |
And we really have to look at each one in turn
link |
and think, why did it only happen once?
link |
In this case, it's really difficult to split water.
link |
It requires a lot of power,
link |
and that power, you're effectively separating charge
link |
across a membrane, and the way in which you do it,
link |
if it doesn't all rush back
link |
and kind of cause an explosion right at the site,
link |
requires really careful wiring.
link |
And that wiring, it can't be easy to get it right
link |
because the plants that we see around us,
link |
they have chloroplasts.
link |
Those chloroplasts were cyanobacteria ones.
link |
Those cyanobacteria are the only group of bacteria
link |
that can do that type of photosynthesis.
link |
So there's plenty of opportunity.
link |
So not even many bacteria.
link |
So who invented photosynthesis?
link |
The cyanobacteria, or their ancestors.
link |
And there's not many?
link |
No other bacteria can do
link |
what's called oxygenic photosynthesis.
link |
Lots of other bacteria can split.
link |
I mean, you can take your hydrogen from somewhere else.
link |
You can take it from hydrogen sulfide
link |
bubbling out of a hydrothermal vent.
link |
Grab your two hydrogens.
link |
The sulfur is the waste now.
link |
You can do it from iron.
link |
You can take electrons.
link |
So the early oceans were probably full of iron.
link |
You can take an electron from ferrous iron,
link |
so iron two plus and make it iron three plus,
link |
which now precipitates as rust,
link |
and you take a proton from the acidic early ocean,
link |
Now you've got a hydrogen atom.
link |
Stick it onto CO2.
link |
You've just done the trick.
link |
The trouble is you bury yourself in rusty iron.
link |
And with sulfur, you can bury yourself in sulfur.
link |
One of the reasons oxygenic photosynthesis
link |
is so much better is that the waste product is oxygen,
link |
which just bubbles away.
link |
That seems like extremely unlikely,
link |
and it's extremely essential
link |
for the evolution of complex organisms
link |
because of all the oxygen.
link |
Yeah, and that didn't accumulate quickly either.
link |
So it's converting, what is it?
link |
It's converting energy from the sun
link |
and the resource of water
link |
into the resource needed for animals.
link |
Both resources needed for animals.
link |
We need to eat, and we need to burn the food,
link |
and we're eating plants,
link |
which are getting their energy from the sun,
link |
and we're burning it with their waste product,
link |
which is the oxygen.
link |
So there's a lot of kind of circularity in that,
link |
but without an oxygenated planet,
link |
you couldn't really have predation.
link |
You can have animals,
link |
but you can't really have animals
link |
that go around and eat each other.
link |
You can't have ecosystems as we know them.
link |
Well, let's actually step back.
link |
What about eukaryotic versus prokaryotic cells, prokaryotes?
link |
What are each of those,
link |
and how big of an invention is that?
link |
I personally think that's the single biggest invention
link |
in the whole history of life.
link |
Yeah, so I mentioned bacteria and archaea.
link |
These are both prokaryotes.
link |
They're basically small cells that don't have a nucleus.
link |
If you look at them under a microscope,
link |
you don't see much going on.
link |
If you look at them under a super resolution microscope,
link |
then they're fantastically complex.
link |
In terms of their molecular machinery, they're amazing.
link |
In terms of their morphological appearance
link |
under a microscope, they're really small and really simple.
link |
The earliest life that we can physically see
link |
on the planet are stromatolites,
link |
which are made by things like cyanobacteria,
link |
and they're large superstructures.
link |
Effectively, biofilms plated on top of each other,
link |
and you end up with quite large structures
link |
that you can see in the fossil record.
link |
But they never came up with animals.
link |
They never came up with plants.
link |
They came up with multicellular things,
link |
filamentous cyanobacteria, for example.
link |
They're just long strings of cells.
link |
But the origin of the eukaryotic cell
link |
seems to have been what's called an endosymbiosis,
link |
so one cell gets inside another cell.
link |
And I think that that's transformed
link |
the energetic possibilities of life.
link |
So what we end up with is a kind of supercharged cell,
link |
which can have a much larger nucleus
link |
with many more genes, all supported.
link |
If you think about it, you could think about it
link |
as multi bacterial power without the overhead.
link |
So you've got a cell and it's got bacteria living in it,
link |
and those bacteria are providing it
link |
with the energy currency it needs.
link |
But each bacterium has a genome of its own,
link |
which costs a fair amount of energy to express,
link |
to kind of turn over and convert into proteins and so on.
link |
What the mitochondria did,
link |
which are these power packs in our own cells,
link |
they were bacteria once,
link |
and they threw away virtually all their genes.
link |
They've only got a few left.
link |
So mitochondria is, like you said,
link |
is the bacteria that got inside a cell
link |
and then throw away all this stuff it doesn't need to,
link |
survive inside the cell, and then kept what?
link |
So what we end up with,
link |
so it kept always a handful of genes.
link |
In our own case, 37 genes.
link |
But there's a few protists, which are single celled things
link |
that have got as many as 70 or 80 genes.
link |
So it's not always the same, but it's always a small number.
link |
And you can think of it as a paired down power pack
link |
where the control unit has really been,
link |
has been kind of paired down to almost nothing.
link |
So you're putting out the same power,
link |
but the investment in the overheads is really paired down.
link |
That means that you can support
link |
a much larger nuclear genome.
link |
So we've gone up in the number of genes,
link |
but also the amount of power you have
link |
to convert those genes into proteins.
link |
We've gone up about fourfold in the number of genes,
link |
but in terms of the size of genomes
link |
and your ability to make the building blocks,
link |
make the proteins, we've gone up 100,000 fold or more.
link |
So it's huge step change in the possibilities of evolution.
link |
And it's interesting then that the only two occasions
link |
that complex life has arisen on Earth,
link |
plants and animals,
link |
fungi you could say are complex as well,
link |
but they don't form such complex morphology
link |
as plants and animals.
link |
Start with a single cell.
link |
They start with an oocyte and a sperm
link |
fused together to make a zygote.
link |
So we start development with a single cell
link |
and all the cells in the organism have identical DNA.
link |
And you switch off in the brain,
link |
you switch off these genes and you switch on those genes
link |
and liver, you switch off those
link |
and you switch on a different set.
link |
And the standard evolutionary explanation for that
link |
is that you're restricting conflict.
link |
You don't have a load of genetically different cells
link |
that are all fighting each other.
link |
The trouble with bacteria, they form these biofilms
link |
and they're all genetically different.
link |
And effectively they're incapable
link |
of that level of cooperation.
link |
They would get in a fight.
link |
Okay, so why is this such a difficult invention
link |
of getting this bacteria inside
link |
and becoming an engine which the mitochondria is?
link |
Why do you assign it such great importance?
link |
Is it great importance in terms of the difficulty
link |
of how it was to achieve or great importance
link |
in terms of the impact it had on life?
link |
It had a huge impact on life
link |
because if that had not happened,
link |
you can be certain that life on earth
link |
would be bacterial only.
link |
And that took a really long time too.
link |
It took 2 billion years.
link |
And it hasn't happened since to the best of our knowledge.
link |
So it looks as if it's genuinely difficult.
link |
And if you think about it then
link |
from just an informational perspective,
link |
you think bacteria have got,
link |
they structure their information differently.
link |
So a bacterial cell has a small genome,
link |
you might have 4,000 genes in it,
link |
but a single E. coli cell has access
link |
to about 30,000 genes potentially.
link |
It's got a kind of metagenome
link |
where other E. coli out there
link |
have got different gene sets
link |
and they can switch them around between themselves.
link |
And so you can generate a huge amount of variation
link |
and they've got more,
link |
an E. coli metagenome is larger than the human genome.
link |
We own 20,000 genes or something.
link |
So, and they've had 4 billion years of evolution
link |
to work out what can I do
link |
and what can't I do with this metagenome?
link |
And the answer is you're stuck, you're still bacteria.
link |
So they have explored genetic sequence space
link |
far more thoroughly than eukaryotes ever did
link |
because they've had twice as long at least
link |
and they've got much larger populations
link |
and they never got around this problem.
link |
So why can't they?
link |
It seems as if you can't solve it with information alone.
link |
So what's the problem?
link |
The problem is structure.
link |
If the very first cells needed an electrical charge
link |
on their membrane to grow and in bacteria,
link |
it's the outer membrane that surrounds the cell
link |
which is electrically charged.
link |
You try and scale that up
link |
and you've got a fundamental design problem,
link |
you've got an engineering problem.
link |
And there are examples of it
link |
and what we see in all these cases
link |
is what's known as extreme polyploidy,
link |
which is to say they have tens of thousands of copies
link |
of their complete genome,
link |
which is energetically hugely expensive
link |
and you end up with a large bacteria
link |
with no further development.
link |
What you need is to incorporate
link |
these electrically charged power pack units inside
link |
with their control units intact
link |
and for them not to conflict so much with the host cell
link |
that it all goes wrong.
link |
Perhaps it goes wrong more often than not.
link |
And then you change the topology of the cell.
link |
Now you don't necessarily have any more DNA
link |
than a giant bacterium with extreme polyploidy,
link |
but what you've got is an asymmetry.
link |
You now have a giant nuclear genome
link |
which surrounded by lots of subsidiary energetic genomes
link |
that do all the, they're the control units
link |
that are doing all the control of energy generation.
link |
Could this have been done gradually
link |
or does it have to be done,
link |
the power pack has to be all intact
link |
and ready to go and working?
link |
I mean, it's a kind of step change
link |
in the possibilities of evolution,
link |
but it doesn't happen overnight.
link |
It's gonna still require multiple, multiple generations.
link |
So it could take millions of years.
link |
It could take shorter times.
link |
There's another thing I would like to put the number of steps
link |
and try and work out what's required at each step.
link |
And we are trying to do that with sex for example.
link |
You can't have a very large genome
link |
unless you have sex at that point.
link |
So what are the changes to go
link |
from bacterial recombination to eukaryotic recombination?
link |
What do you need to do?
link |
Why do we go from passing around bits of DNA
link |
as if it's loose change to fusing cells together,
link |
lining up the chromosomes,
link |
recombining across the chromosomes,
link |
and then going through two rounds of cell division
link |
to produce your gametes?
link |
All eukaryotes do it that way.
link |
So again, why switch?
link |
What are the drivers here?
link |
So there's a lot of time, there's a lot of evolution,
link |
but as soon as you've got cells living inside another cell,
link |
what you've got is a new design.
link |
You've got new potential that you didn't have before.
link |
So the cell living inside another cell, that design
link |
allows for better storage of information,
link |
better use of energy, more delegation,
link |
like a hierarchical control of the whole thing.
link |
And then somehow that leads to ability
link |
to have multi cell organisms.
link |
I'm not sure that you have hierarchical control necessarily,
link |
but you've got a system where you can have
link |
a much larger information storage depot in the nucleus.
link |
You can have a much larger genome.
link |
And that allows multicellularity, yes,
link |
because it allows you, it's a funny thing,
link |
to have an animal where I have 70% of my genes
link |
switched on in my brain,
link |
and a different 50% switched on in my liver or something,
link |
you've got to have all those genes in the egg cell
link |
at the very beginning,
link |
and you've got to have a program of development
link |
which says, okay, you guys switch off those genes
link |
and switch on those genes, and you guys, you do that.
link |
But all the genes are there at the beginning.
link |
That means you've got to have a lot of genes in one cell
link |
and you've got to be able to maintain them.
link |
And the problem with bacteria is they don't get close
link |
to having enough genes in one cell.
link |
So if you were to try and make a multicellular organism
link |
from bacteria, you'd bring different types
link |
of bacteria together and hope they'll cooperate.
link |
And the reality is they don't.
link |
That's really, really tough to do.
link |
We know they don't because it doesn't exist.
link |
We have the data as far as we know.
link |
I'm sure there's a few special ones
link |
and they dead off quickly.
link |
I'd love to know some of the most fun things
link |
bacteria have done since.
link |
Oh, there's a few.
link |
I mean, they can do some pretty funky things.
link |
And this is broad brushstroke that I'm talking about.
link |
Generally speaking.
link |
So how was, so another fun invention.
link |
Us humans seem to utilize it well,
link |
but you say it's also very important early on is sex.
link |
Just asking for a friend.
link |
And when was it invented and how hard was it to invent,
link |
just as you were saying, and why was it invented?
link |
Why, how hard was it and when?
link |
I have a PhD student who's been working on this
link |
and we've just published a couple of papers on sex.
link |
What do you publish?
link |
Does biology, is it biology, genetics, journals?
link |
This is actually PNAS,
link |
which is Proceedings of the National Academy.
link |
Broad, big, big picture stuff.
link |
Everyone's interested in sex.
link |
And the job of a biologist is to make sex dull.
link |
Yes, yeah, that's a beautiful way to put it.
link |
Okay, so when was it invented?
link |
It was invented with eukaryotes about two billion years ago.
link |
All eukaryotes share the same basic mechanism
link |
that you produce gametes, the gametes fuse together.
link |
So a gamete is the egg cell and the sperm.
link |
They're not necessarily even different in size or shape.
link |
So the simplest eukaryotes produce
link |
what are called motile gametes.
link |
They're all like sperm and they all swim around.
link |
They find each other, they fuse together.
link |
They don't have kind of much going on there beyond that.
link |
And then these are haploid,
link |
which is to say we all have two copies of our genome
link |
and the gametes have only a single copy of the genome.
link |
So when they fuse together, you now become diploid again,
link |
which is to say you now have two copies of your genome.
link |
And what you do is you line them all up
link |
and then you double everything.
link |
So now we have four copies of the complete genome.
link |
And then we crisscross between all of these things.
link |
So we take a bit from here and stick it on there
link |
and a bit from here and we stick it on here.
link |
That's recombination.
link |
And then we go through two rounds of cell division.
link |
So we divide in half.
link |
So now the two daughter cells have two copies
link |
and we divide in half again.
link |
Now we have some gametes,
link |
each of which has got a single copy of the genome.
link |
And that's the basic ground plan
link |
for what's called meiosis and Syngami.
link |
That's basically sex.
link |
And it happens at the level of single celled organisms.
link |
And it happens pretty much the same way in plants
link |
and pretty much the same way in animals and so on.
link |
And it's not found in any bacteria.
link |
They switch things around using the same machinery
link |
and they take up a bit of DNA from the environment.
link |
They take out this bit and stick in that bit
link |
and it's the same molecular machinery they're using to do it.
link |
So what about the kind of, you said, find each other,
link |
this kind of imperative, find each other.
link |
Well, you've got a few cells together.
link |
So the bottom line on all of this is bacteria.
link |
I mean, it's kind of simple when you've figured it out
link |
and figuring it out, this is not me,
link |
this is my PhD student, Marco Colnaghi.
link |
And in effect, if you're doing lateral,
link |
you're a Nicoli cell, you've got 4,000 genes.
link |
You wanna scale up to a eukaryotic size.
link |
I wanna have 20,000 genes.
link |
And I need to maintain my genome
link |
so it doesn't get shot to pieces by mutations.
link |
And I'm gonna do it by lateral gene transfer.
link |
So I know I've got a mutation in a gene.
link |
I don't know which gene it is because I'm not sentient,
link |
but I know I can't grow.
link |
I know all my regulation systems are saying,
link |
something wrong here, something wrong, pick up some DNA,
link |
pick up a bit of DNA from the environment.
link |
If you've got a small genome,
link |
the chances of you picking up the right bit of DNA
link |
from the environment is much higher
link |
than if you've got a genome of 20,000 genes.
link |
To do that, you've effectively got to be picking up DNA
link |
all the time, all day long and nothing else.
link |
And you're still gonna get the wrong DNA.
link |
You've got to pick up large chunks.
link |
And in the end, you've got to align them.
link |
You're forced into sex, to coin a phrase.
link |
So there is a kind of incentive.
link |
If you wanna have a large genome,
link |
you've got to prevent it mutating to nothing.
link |
That will happen with bacteria.
link |
This is another reason why bacteria
link |
can't have a large genome.
link |
But as soon as you give them the power pack,
link |
as soon as you give eukaryotic cells the power pack
link |
that allows them to increase the size of their genome,
link |
then you face the pressure
link |
that you've got to maintain its quality.
link |
You've got to stop it just mutating away.
link |
What about sexual selection?
link |
So the finding, like, I don't like this one.
link |
I don't like this one.
link |
This one seems all right.
link |
Like, what's the...
link |
At which point does it become less random?
link |
It's hard to know.
link |
Because eukaryotes just kind of float around.
link |
Just kind of have...
link |
Yeah, I mean, is there sexual selection
link |
in single celled eukaryotes?
link |
There probably is.
link |
It's just that I don't know very much about it.
link |
By the time we get onto...
link |
You don't hang out with the eukaryotes.
link |
Well, I do all the time, but...
link |
But you can't communicate with them yet.
link |
Yeah, a peacock or something.
link |
The kind of standard answer,
link |
this is not quite what I work on,
link |
but the standard answer is that it's female mate choice.
link |
She is looking for good genes.
link |
And if you can have a tail that's like this
link |
and still survive, still be alive,
link |
not actually have been taken down by the nearest predator,
link |
then you must've got pretty good genes
link |
because despite this handicap, you're able to survive.
link |
So those are like human interpretable things,
link |
like with a peacock.
link |
But I wonder, I'm sure echoes of the same thing
link |
are there with more primitive organisms.
link |
Basically your PR, like how you advertise yourself
link |
that you're worthy of.
link |
So one big advertisement is the fact
link |
that you survived it all.
link |
Let me give you one beautiful example of an algal bloom.
link |
And this can be a sign of bacteria.
link |
It's gonna be in bacteria.
link |
So if suddenly you pump nitrate or phosphate
link |
or something into the ocean and everything goes green,
link |
you end up with all this algae growing there.
link |
A viral infection or something like that
link |
can kill the entire bloom overnight.
link |
And it's not that the virus takes out everything overnight.
link |
It's that most of the cells in that bloom kill themselves
link |
before the virus can get onto them.
link |
And it's through a form of cell death
link |
called programmed cell death.
link |
And we do the same things.
link |
It's how we have the gaps between our fingers and so on.
link |
It's how we craft synapses in the brain.
link |
It's fundamental again to multicellular life.
link |
They have the same machinery in these algal blooms.
link |
How do they know who dies?
link |
The answer is they will often put out a toxin.
link |
And that toxin is kind of a challenge to you.
link |
Either you can cope with the toxin or you can't.
link |
If you can cope with it, you form a spore
link |
and you will go on to become the next generation.
link |
You're forming kind of a resistance spore.
link |
You sink down a little bit, you get out of the way,
link |
you're out of the, you can't be attacked by a virus
link |
if you're a spore or it's not so easily.
link |
Whereas if you can't deal with that toxin,
link |
you pull the plug and you trigger your death apparatus
link |
and you kill yourself.
link |
Oh, so it's truly life and death selection.
link |
Yeah, so it's really, it's a challenge.
link |
And this is a bit like sexual selection.
link |
It's not so, they're all pretty much genetically identical,
link |
but they've had different life histories.
link |
So have you had a tough day?
link |
Did you happen to get infected by this virus?
link |
Or did you run out of iron?
link |
Or did you get a bit too much sun?
link |
Whatever it may be, if this extra stress of the toxin
link |
just pushes you over the edge,
link |
then you have this binary choice.
link |
Either you're the next generation
link |
or you kill yourself now using this same machinery.
link |
It's also actually exactly the way I approach dating,
link |
but that's probably why I'm single.
link |
Okay, what about if we can step back, DNA?
link |
Just mechanism of storing information.
link |
RNA, DNA, how big of an invention was that?
link |
That seems to be, that seems to be fundamental
link |
to like something deep within what life is,
link |
is the ability, as you said,
link |
to kind of store and propagate information.
link |
But then you also kind of infer that
link |
with your and your students work,
link |
that there's a deep connection between the chemistry
link |
and the ability to have this kind of genetic information.
link |
So how big of an invention is it
link |
to have a nice representation,
link |
nice hard drive for info to pass on?
link |
I mean, but when I was talking about the code,
link |
you see the code in RNA as well.
link |
And RNA almost certainly came first.
link |
And there's been an idea going back decades
link |
called the RNA world,
link |
because RNA in theory can copy itself
link |
and can catalyze reactions.
link |
So it kind of cuts out this chicken and egg loop.
link |
So DNA as possible is not that special.
link |
So RNA, RNA is the thing that does the work really.
link |
And the code lies in RNA.
link |
The code lies in the interactions
link |
between RNA and amino acids.
link |
And it still is there today in the ribosome, for example,
link |
which is just kind of a giant ribozyme,
link |
which is to say it's an enzyme that's made of RNA.
link |
So getting to RNA, I suspect is probably not that hard,
link |
but getting from RNA, how do you,
link |
you know, there's multiple different types of RNA now.
link |
How do you distinguish?
link |
This is something we're actively thinking about.
link |
How do you distinguish between,
link |
you know, a random population of RNA?
link |
Some of them go on to become messenger RNA.
link |
This is the transcript of the code
link |
of the gene that you want to make.
link |
Some of them become transfer RNA,
link |
which is kind of the unit that holds the amino acid
link |
that's going to be polymerized.
link |
Some of them become ribosomal RNA,
link |
which is the machine which is joining them all up together.
link |
How do they discriminate themselves?
link |
And, you know, is some kind of phase transition
link |
It's a difficult question.
link |
And we're now in the region of biology
link |
where information is coming in.
link |
But the thing about RNA is very, very good at what it does.
link |
But the largest genome supported by RNA
link |
are RNA viruses like HIV, for example.
link |
They're pretty small.
link |
And so there's a limit to how complex life could be
link |
unless you come up with DNA,
link |
which chemically is a really small change.
link |
But how easy it is to make that change,
link |
I don't really know.
link |
As soon as you've got DNA,
link |
then you've got an amazingly stable molecule
link |
for information storage.
link |
And you can do absolutely anything.
link |
But how likely that transition from RNA to DNA was,
link |
I don't know either.
link |
How much possibility is there for variety
link |
in ways to store information?
link |
Because it seems to be very,
link |
there's specific characteristics
link |
about the programming language of DNA.
link |
Yeah, there's a lot of work going on
link |
on what's called the xenodNA or RNA.
link |
Can we replace the bases themselves,
link |
the letters, if you like, in RNA or DNA?
link |
Can we replace the backbone?
link |
Can we replace, for example, phosphate with arsenate?
link |
Can we replace the sugar ribose or deoxyribose
link |
with a different sugar?
link |
And the answer is yes, you can.
link |
Within limits, there's not an infinite space there.
link |
Arsenate doesn't really work
link |
if the bonds are not as strong as phosphate.
link |
It's probably quite hard to replace phosphate.
link |
It's possible to do it.
link |
The question to me is why is it this way?
link |
Is it because there was some form of selection
link |
that this is better than the other forms
link |
and there were lots of competing forms
link |
of information storage early on
link |
and this one was the one that worked out?
link |
Or was it kind of channeled that way,
link |
that these are the molecules that you're dealing with
link |
And I'm increasingly thinking it's that way,
link |
that we're channeled towards ribose, phosphate,
link |
and the bases that are used.
link |
But there are 200 different letters
link |
kicking around out there that could have been used.
link |
It's such an interesting question.
link |
If you look in the programming world in computer science,
link |
there's a programming language called JavaScript,
link |
which was written super quickly.
link |
It's a giant mess, but it took over the world.
link |
And it was kind of a...
link |
Sounds very biological.
link |
It was kind of a running joke that like,
link |
like surely this can't be,
link |
this is a terrible programming language.
link |
It's a giant mess.
link |
It's full of bugs.
link |
It's so easy to write really crappy code,
link |
but it took over all a front end development
link |
in the web browser.
link |
If you have any kind of dynamic interactive website,
link |
it's usually running JavaScript.
link |
And it's now taking over much of the backend,
link |
which is like the serious heavy duty computational stuff.
link |
And it's become super fast
link |
with the different compilation engines that are running it.
link |
So it's like, it really took over the world.
link |
It's very possible that this initially crappy derided language
link |
actually takes everything over.
link |
And then the question is,
link |
did human civilization always strive towards JavaScript?
link |
Or was JavaScript just the first programming language
link |
that ran on the browser and still sticky?
link |
The first is the sticky one.
link |
And so it wins over anything else because it was first.
link |
And I don't think that's answerable, right?
link |
But it's good to ask that.
link |
I suppose in the lab,
link |
you can't run it with programming languages,
link |
but in biology you can probably do some kind of
link |
small scale evolutionary test to try to infer,
link |
I mean, in a way we've got the hardware
link |
and the software here.
link |
And the hardware is maybe the DNA and the RNA itself.
link |
And then the software perhaps is more about the code.
link |
Did the code have to be this way?
link |
Could it have been a different way?
link |
People talk about the optimization of the code
link |
and there's some suggestion for that.
link |
I think it's weak actually.
link |
But you could imagine you could come out
link |
with a million different codes
link |
and this would be one of the best ones.
link |
Well, we don't know this.
link |
Well, I mean, people have tried to model it
link |
based on the effect that mutations would have.
link |
So no, you're right.
link |
We don't know because that's a single assumption
link |
that a mutation is what's being selected on there.
link |
And there's other possibilities too.
link |
I mean, there does seem to be a resilience
link |
and a redundancy to the whole thing.
link |
It's hard to mess up and the way you mess it up
link |
often is likely to produce interesting results.
link |
Are you talking about JavaScript or the genetic code now?
link |
Yeah, well, I mean, it's almost,
link |
biology is underpinned by this kind of mess as well.
link |
And you look at the human genome and it's full of stuff
link |
that is really either broken or dysfunctional
link |
or was a virus once, whatever it may be.
link |
And somehow it works.
link |
And maybe we need a lot of this mess.
link |
We know that some functional genes are taken from this mess.
link |
So what about, you mentioned the predatory behavior.
link |
We talked about sex.
link |
What about violence, predator and prey dynamics?
link |
When was that invented?
link |
And poetic and biological ways of putting it,
link |
how do you describe predator prey relationship?
link |
Is it a beautiful dance or is it a violent atrocity?
link |
Well, I guess it's both, isn't it?
link |
I mean, when does it start?
link |
It starts in bacteria.
link |
You see these amazing predators.
link |
Della Vibrio is one that Lynn Margulis
link |
used to talk about a lot.
link |
It's got a kind of a drill piece
link |
that drills through the wall
link |
and the membrane of the bacterium.
link |
And then it effectively eats the bacterium
link |
from just inside the periplasmic space
link |
and makes copies of itself that way.
link |
So that's straight predation.
link |
There are predators among bacteria.
link |
So predation in that, sorry to interrupt,
link |
means you murder somebody
link |
and use their body as a resource in some way.
link |
But it's not parasitic in that
link |
you need them to be still alive.
link |
No, no, I mean, predation is you kill them, really.
link |
Parasites, so you kind of live on them.
link |
Okay, so, but it seems the predator is the really popular.
link |
So what we see if we go back 560, 570 million years
link |
before the Cambrian explosion,
link |
there is what's known as the Ediacaran fauna,
link |
or sometimes they call Vendobionts,
link |
which is a lovely name.
link |
And it's not obvious that they're animals at all.
link |
They're stalked things.
link |
They often have fronds that look a lot like leaves
link |
with kind of fractal branching patterns on them.
link |
And the thing is, they're found,
link |
sometimes geologists can figure out the environment
link |
that they were in and say,
link |
this is more than 200 meters deep
link |
because there's no sign of any waves.
link |
There's no storm damage down here, this kind of thing.
link |
They were more than 200 meters deep,
link |
so they're definitely not photosynthetic.
link |
These are animals and they're filter feeders.
link |
And we know sponges and corals and things
link |
are filter feeding animals.
link |
They're stuck to the spot.
link |
And little bits of carbon that come their way,
link |
they filter it out and that's what they're eating.
link |
So no predation involved in this,
link |
beyond stuff just dies anyway.
link |
And it feels like a very gentle, rather beautiful,
link |
rather limited world, you might say.
link |
There's not a lot going on there.
link |
And something changes.
link |
Oxygen definitely changes during this period.
link |
Other things may have changed as well.
link |
But the next thing you really see in the fossil record
link |
is the Cambrian explosion.
link |
And what do we see there?
link |
We're now seeing animals that we would recognize.
link |
They've got eyes, they've got claws, they've got shells.
link |
They're plainly killing things or running away and hiding.
link |
And so we've gone from a rather gentle but limited world
link |
to a rather vicious, unpleasant world that we recognize
link |
and which leads to kind of arms races,
link |
evolutionary arms races, which again is something
link |
that when we think about a nuclear arms race,
link |
we think, Jesus, we don't want to go there.
link |
It's not done anybody any good.
link |
In some ways, maybe it does do good.
link |
I don't want to make an argument for nuclear arms.
link |
But predation as a mechanism forces organisms
link |
to adapt to change to be better to escape or to kill.
link |
If you need to eat, then you've got to eat.
link |
And a cheetah's not going to run at that speed
link |
unless it has to because the zebra is capable of escaping.
link |
So it leads to much greater feats of evolution
link |
than would ever have been possible without it.
link |
And in the end, to a much more beautiful world.
link |
And so it's not all bad by any means.
link |
But the thing is you can't have this
link |
if you don't have an oxygenated planet.
link |
Because it's all in the end, it's about how much energy
link |
can you extract from the food you eat.
link |
And if you don't have an oxygenated planet,
link |
you can get about 10% out, not much more than that.
link |
And if you've got an oxygenated planet,
link |
you can get about 40% out.
link |
And that means you can have,
link |
instead of having one or two trophic levels,
link |
you can have five or six trophic levels.
link |
And that means things can eat things
link |
that eat other things and so on.
link |
And you've gone to a level of ecological complexity,
link |
which is completely impossible in the absence of oxygen.
link |
This reminds me of the Hunter S. Thompson quote,
link |
that for every moment of triumph,
link |
for every instance of beauty, many souls must be trampled.
link |
The history of life on Earth, unfortunately,
link |
is that of violence.
link |
Just the trillions and trillions of multi cell organisms
link |
that were murdered in the struggle for survival.
link |
It's a sorry statement, but yes, it's basically true.
link |
And that somehow is a catalyst
link |
from an evolutionary perspective for creativity,
link |
for creating more and more complex organisms
link |
that are better and better at surviving.
link |
I mean, survival of the fittest,
link |
if you just go back to that old phrase,
link |
means death of the weakest.
link |
Now, what's fit, what's weak,
link |
these are terms that don't have much intrinsic meaning.
link |
But the thing is, evolution only happens because of death.
link |
One way to die is the constraints,
link |
the scarcity of the resources in the environment,
link |
but that seems to be not nearly as good of a mechanism
link |
for death than other creatures
link |
roaming about in the environment.
link |
When I say environment, I mean like the static environment,
link |
but then there's the dynamic environment
link |
of bigger things trying to eat you
link |
and use you for your energy.
link |
It forces you to come up with a solution
link |
to your specific problem that is inventive
link |
and is new and hasn't been done before.
link |
And so it forces, I mean, literally change,
link |
literally evolution on populations.
link |
They have to become different.
link |
And it's interesting that humans have channeled that
link |
into more, I mean, I guess what humans are doing
link |
is they're inventing more productive
link |
and safe ways of doing that.
link |
You know, this whole idea of morality
link |
and all those kinds of things,
link |
I think they ultimately lead to competition
link |
versus violence, because I think violence
link |
can have a cold, brutal, inefficient aspect to it.
link |
But if you channel that into more controlled competition
link |
in the space of ideas, in the space of approaches to life,
link |
maybe you can be even more productive than evolution is.
link |
Because evolution is very wasteful.
link |
Like the amount of murder required
link |
to really test a good idea,
link |
genetically speaking, is just a lot.
link |
Many, many, many generations.
link |
Morally, we cannot base society
link |
on the way that evolution works.
link |
That's an invention, right?
link |
But actually, in some respects we do,
link |
which is to say, this is how science works.
link |
We have competing hypotheses that have to get better,
link |
otherwise they die.
link |
It's the way that society works.
link |
You know, in ancient Greece, we had the Athens
link |
and Sparta and city states,
link |
and then we had the Renaissance and nation states,
link |
and universities compete with each other.
link |
Tremendous amount of companies competing
link |
with each other all the time.
link |
It drives innovation.
link |
And if we want to do it without all the death
link |
that we see in nature,
link |
then we have to have some kind of societal level control
link |
that says, well, there's some limits, guys,
link |
and these are what the limits are gonna be.
link |
And society as a whole has to say,
link |
right, we want to limit the amount of death here,
link |
so you can't do this and you can't do that.
link |
And you know, who makes up these rules,
link |
and how do we know?
link |
It's a tough thing, but it's basically
link |
trying to find a moral basis
link |
for avoiding the death of evolution and natural selection
link |
and keeping the innovation and the richness of it.
link |
And I forgot who said it, but that murder is illegal.
link |
Probably Kurt Vonnegut.
link |
Murder is illegal except when it's done
link |
to the sound of trumpets and at a large scale.
link |
So we still have wars.
link |
But we are struggling with this idea
link |
that murder is a bad thing.
link |
It's so interesting how we're channeling
link |
the best of the evolutionary imperative
link |
and trying to get rid of the stuff that's not productive.
link |
Trying to almost accelerate evolution.
link |
The same kind of thing that makes evolution creative.
link |
We're trying to use that.
link |
I think we naturally do it.
link |
I mean, I don't think we can help ourselves do it.
link |
And you know, capitalism as a form
link |
is basically about competition and differential rewards.
link |
But we, society, and you know, we have a,
link |
I keep using this word moral obligation,
link |
but you know, we cannot operate as a society
link |
if we go that way.
link |
It's interesting that we've had problems achieving balance.
link |
So for example, in the financial crash in 2009,
link |
do you let banks go to the wall or not?
link |
This kind of question.
link |
In evolution, certainly you let them go to the wall.
link |
And in that sense, you don't need the regulation
link |
because they just die.
link |
Whereas if we, as a society,
link |
think about what's required for society as a whole,
link |
then you don't necessarily let them go to the wall.
link |
In which case you then have to impose
link |
some kind of regulation that the bankers themselves will,
link |
in an evolutionary manner, exploit.
link |
Yeah, we've been struggling with this kind of idea
link |
of capitalism, the cold brutality of capitalism
link |
that seems to create so much beautiful things
link |
And then the ideals of communism
link |
that seem to create so much brutal destruction in history.
link |
We struggle with ideas of,
link |
well, maybe we didn't do it right.
link |
How can we do things better?
link |
And then the ideas are the things
link |
where we're playing with as opposed to people.
link |
If a PhD student has a bad idea,
link |
we don't shoot the PhD student.
link |
We just criticize their idea and hope they improve.
link |
You have a very humane lab.
link |
Yeah, I don't know how you guys do it.
link |
The way I run things, it's always life and death.
link |
Okay, so it is interesting about humans
link |
that there is an inner sense of morality
link |
which begs the question of how did homo sapiens evolve?
link |
If we think about the invention of,
link |
early invention of sex and early invention of predation,
link |
what was the thing invented to make humans?
link |
What would you say?
link |
I mean, I suppose a couple of things I'd say.
link |
Number one is you don't have to wind the clock back
link |
very far, five, six million years or so,
link |
and let it run forwards again.
link |
And the chances of humans as we know them
link |
is not necessarily that high.
link |
You know, imagine as an alien, you find planet Earth
link |
and it's got everything apart from humans on it.
link |
It's an amazing, wonderful, marvelous planet,
link |
but nothing that we would recognize
link |
as extremely intelligent life,
link |
kind of space faring civilization.
link |
So when we think about aliens,
link |
we're kind of after something like ourselves.
link |
We're after a space faring civilization.
link |
We're not after zebras and giraffes and lions and things,
link |
amazing though they are.
link |
But the additional kind of evolutionary steps
link |
to go from large, complex mammals, monkeys, let's say,
link |
to humans doesn't strike me as that long a distance.
link |
It's all about the brain.
link |
And where's the brain and morality coming from?
link |
It seems to me to be all about groups,
link |
human groups and interactions between groups.
link |
The collective intelligence of it.
link |
Yes, the interactions really.
link |
And there's a guy at UCL called Mark Thomas,
link |
who's done a lot of really beautiful work,
link |
I think, on this kind of question.
link |
So I talk to him every now and then,
link |
so my views are influenced by him.
link |
But a lot seems to depend on population density,
link |
that the more interactions you have going on
link |
between different groups, the more transfer of information,
link |
if you like, between groups,
link |
people moving from one group to another group,
link |
almost like lateral gene transfer in bacteria,
link |
the more expertise you're able to develop and maintain,
link |
the more culturally complex your society can become.
link |
And groups that have become detached,
link |
like on Easter Island, for example,
link |
very often degenerate in terms of the complexity
link |
of their civilization.
link |
Is that true for complex organisms in general?
link |
Population density is often productive.
link |
Really matters, but in human terms,
link |
I don't know what the actual factors were
link |
that were driving a large brain,
link |
but you can talk about fire, you can talk about tool use,
link |
you can talk about language,
link |
and none of them seem to correlate especially well
link |
with the actual known trajectory of human evolution
link |
in terms of cave art and these kind of things.
link |
That seems to work much better
link |
just with population density
link |
and number of interactions between different groups,
link |
all of which is really about human interactions,
link |
human human interactions and the complexity of those.
link |
But population density is the thing
link |
that increases the number of interactions,
link |
but then there must have been inventions
link |
forced by that number of interactions
link |
that actually led to humans.
link |
So like Richard Wrangham talks about that
link |
it's basically the beta males had to beat up the alpha male.
link |
So that's what collaboration looks like,
link |
is they, when you're living together,
link |
our early ancestors don't like the dictatorial aspect
link |
of a single individual at the top of a tribe.
link |
So they learned to collaborate
link |
how to basically create a democracy of sorts,
link |
a democracy that prevents, minimizes,
link |
or lessens the amount of violence,
link |
which essentially gives strength to the tribe
link |
and make the war between tribes versus the dictator.
link |
I mean, I think one of the most wonderful things
link |
about humans is we're all of those things.
link |
I mean, we are deeply social as a species
link |
and we're also deeply selfish.
link |
And it seems to me the conflict
link |
between capitalism and communism,
link |
it's really just two aspects of human nature,
link |
both of which are.
link |
We have both and we have a constant kind of vying
link |
between the two sides.
link |
We really do care about other people beyond our families,
link |
beyond our immediate people.
link |
We care about society and the society that we live in.
link |
And you could say that's a drawing
link |
towards socialism or communism.
link |
On the other side, we really do care about ourselves.
link |
We really do care about our families,
link |
about working for something that we gain from.
link |
And that's the capitalist side of it.
link |
They're both really deeply ingrained in human nature.
link |
In terms of violence and interactions between groups,
link |
yes, all this dynamic of,
link |
if you're interacting between groups,
link |
you can be certain that they're gonna be burning each other
link |
and all kinds of physical violent interactions as well,
link |
which will drive the kind of cleverness
link |
of how do you resist this?
link |
Let's build a tower.
link |
What are we gonna do to prevent being overrun
link |
by those marauding gangs from over there?
link |
And you look outside humans
link |
and you look at chimps and bonobos and so on,
link |
and they're very, very different structures to society.
link |
Chimps tend to have an aggressive alpha male type structure
link |
and bonobos, there's basically a female society
link |
where the males are predominantly excluded
link |
and only brought in at the behest of the female.
link |
We have a lot in common with both of those groups.
link |
And there's, again, tension there.
link |
And probably chimps, more violence,
link |
the bonobos, probably more sex.
link |
That's another tension.
link |
How serious do we wanna be?
link |
How much fun we wanna be?
link |
Asking for a friend again,
link |
what do you think happened to Neanderthals?
link |
What did we cheeky humans do to the Neanderthals,
link |
Do you think we murdered them?
link |
Was it, how do we murder them?
link |
How do we outcompete them?
link |
Or do we mate them?
link |
I mean, I think there's unequivocal evidence
link |
that we mated with them.
link |
We always try to mate with everything.
link |
There's some interesting,
link |
the first sequences that came along
link |
were in mitochondrial DNA.
link |
And that was back to about 2002 or thereabouts.
link |
What was found was that Neanderthal mitochondrial DNA
link |
was very different to human mitochondria.
link |
Oh, that's so interesting.
link |
You could do a clock on it
link |
and it said the divergent state
link |
was about 600,000 years ago or something like that.
link |
So not so long ago.
link |
And then the first full genomes were sequenced
link |
maybe 10 years after that.
link |
And they showed plenty of signs of mating between.
link |
So the mitochondrial DNA effectively says no mating.
link |
And the nuclear genes say, yeah, lots of mating.
link |
But we don't know.
link |
How's that possible?
link |
So can you explain the difference
link |
between mitochondrial DNA and nucleus?
link |
I've talked before about the mitochondria,
link |
which are the power packs in cells.
link |
These are the paired down control units is their DNA.
link |
So it's passed on by the mother only.
link |
And in the egg cell,
link |
we might have half a million copies of mitochondrial DNA.
link |
There's only 37 genes left and they do a,
link |
it's basically the control unit of energy production.
link |
That's what it's doing.
link |
It's a basic old school machine that does.
link |
And it's got genes that were considered
link |
to be effectively trivial
link |
because they did a very narrowly defined job,
link |
but they're not trivial in the sense
link |
that that narrowly defined job
link |
is about everything that is being alive.
link |
So they're much easier to sequence.
link |
You've got many more copies of these things
link |
and you can sequence them very quickly.
link |
But the problem is because they go down
link |
only the maternal line from mother to daughter,
link |
your mitochondrial DNA and mine is going nowhere.
link |
It doesn't matter.
link |
Any kids we have, they get their mother's mitochondrial DNA
link |
except in very, very rare and strange circumstances.
link |
And so it tells a different story
link |
and it's not a story which is easy to reconcile always.
link |
And what it seems to suggest to my mind at least
link |
is that there was one way traffic of genes
link |
probably going from humans into Neanderthals
link |
rather than the other way around.
link |
Why did the Neanderthals disappear?
link |
I mean, I suspect that they were,
link |
I suspect they were probably less violent,
link |
less clever, less populous, less willing to fight.
link |
I mean, I think it probably drove them to extinction
link |
at the margins of Europe.
link |
And it's interesting how much,
link |
if we ran Earth over and over again,
link |
how many of these branches of intelligent beings
link |
that have figured out some kind of
link |
how to leverage collective intelligence,
link |
which ones of them emerge?
link |
Which ones of them succeed?
link |
Is it the more violent ones?
link |
Is it the more isolated ones?
link |
Like what dynamics result in more productivity?
link |
And I suppose we'll never know.
link |
The more complex the organism,
link |
the harder it is to run the experiment in the lab.
link |
And in some respects, maybe it's best if we don't know.
link |
The truth might be very painful.
link |
What about if we actually step back
link |
a couple of interesting things that we humans do?
link |
One is object manipulation and movement.
link |
And of course, movement was something that was done,
link |
that was another big invention,
link |
being able to move around the environment.
link |
And the other one is this sensory mechanism,
link |
how we sense the environment.
link |
One of the coolest high definition ones is vision.
link |
How big are those inventions
link |
in the history of life on Earth?
link |
Vision, movement, I mean, again, extremely important,
link |
going back to the origin of animals,
link |
the Cambrian explosion where suddenly you're seeing eyes
link |
in the fossil record.
link |
And you can, it's not necessarily, again,
link |
lots of people historically have said
link |
what use is half an eye?
link |
And you can go in a series of steps
link |
from a light sensitive spot on a flat piece of tissue
link |
to an eyeball with a lens and so on.
link |
If you assume no more than, I don't remember,
link |
this was a specific model that I have in mind,
link |
but it was 1% change or half a percent change
link |
for each generation.
link |
How long would it take to evolve an eye as we know it?
link |
And the answer is half a million years.
link |
It doesn't have to take long.
link |
That's not how evolution works.
link |
That's not an answer to the question.
link |
It just shows you can reconstruct the steps
link |
and you can work out roughly how it can work.
link |
So it's not that big a deal to evolve an eye,
link |
but once you have one, then there's nowhere to hide.
link |
And again, we're back to predator prey relationships.
link |
We're back to all the benefits
link |
that being able to see brings you.
link |
And if you think philosophically what bats are doing
link |
with eco location and so on, I have no idea,
link |
but I suspect that they form an image of the world
link |
in pretty much the same way that we do.
link |
It's just a matter of mental reconstruction.
link |
So I suppose the other thing about sight,
link |
there are single celled organisms that have got a lens
link |
and a retina and a cornea and so on.
link |
Basically they've got a camera type eye in a single cell.
link |
They don't have a brain.
link |
What they understand about their world
link |
is impossible to say, but they're capable of coming up
link |
with the same structures to do so.
link |
So I suppose then is that once you've got things like eyes,
link |
then you have a big driving pressure
link |
on the central nervous system
link |
to figure out what it all means.
link |
And then we come around to your other point
link |
about manipulation, sensory input, and so on
link |
about now you have a huge requirement
link |
to understand what your environment is and what it means
link |
and how it reacts and how you should run away
link |
and where you should stay put.
link |
Actually on that point, let me,
link |
I don't know if you know the work of Donald Hoffman,
link |
who uses the argument, the mechanism of evolution
link |
to say that there's not necessarily
link |
a strong evolutionary value to seeing the world as it is.
link |
So objective reality, that our perception actually
link |
is very different from what's objectively real.
link |
We're living inside an illusion
link |
and we're basically the entire set of species on earth,
link |
I think, I guess, are competing in a space
link |
that's an illusion that's distinct from,
link |
that's far away from physical reality as it is,
link |
as defined by physics.
link |
I'm not sure it's an illusion so much as a bubble.
link |
I mean, we have a sensory input,
link |
which is a fraction of what we could have
link |
a sensory input on, and we interpret it
link |
in terms of what's useful for us to know to stay alive.
link |
So yes, it's an illusion in that sense,
link |
but the tree is physically there.
link |
And if you walk into that tree, you know,
link |
that there is, it's not purely a delusion,
link |
there's some physical reality to it.
link |
So it's a sensory slice into reality as it is,
link |
but because it's just a slice,
link |
you're missing a big picture.
link |
But he says that that slice doesn't necessarily
link |
need to be a slice.
link |
It could be a complete fabrication
link |
that's just consistent amongst the species,
link |
which is an interesting, or at least it's a humbling
link |
realization that our perception is limited
link |
and our cognitive abilities are limited.
link |
And at least to me, it's argument from evolution,
link |
I don't know how much, how strong that is as an argument,
link |
but I do think that life can exist in the mind.
link |
In the same way that you can do a virtual reality video game
link |
and you can have a vibrant life inside that place
link |
and that place is not real in some sense,
link |
but you could still have a vibrant,
link |
all the same forces of evolution,
link |
all the same competition, the dynamics of between humans
link |
you can have, but I don't know if,
link |
I don't know if there's evidence for that being
link |
the thing that happened on earth.
link |
It seems that earth.
link |
I think in either environment, I wouldn't deny
link |
that you could have exactly the world that you talk about
link |
and it would be very difficult to,
link |
the idea in matrix movies and so on
link |
that the whole world is completely a construction
link |
and we're fundamentally deluded.
link |
It's difficult to say that's impossible or couldn't happen
link |
or, and certainly we construct in our minds
link |
what the outside world is, but we do it on input
link |
and that input, I would hesitate to say it's not real
link |
because it's precisely how we do understand the world.
link |
We have eyes, but if you keep someone in,
link |
apparently this kind of thing happens,
link |
someone kept in a dark room for five years
link |
or something like that, they never see properly again
link |
because the neural wiring that underpins
link |
how we interpret vision never developed.
link |
You need, when you watch a child develop,
link |
it walks into a table, it bangs his head on the table
link |
and it hurts and now you've got two inputs.
link |
You've got one pain from this sharp edge
link |
and number two, you probably, you've touched it
link |
and realized it's there, it's a sharp edge
link |
and you've got the visual input
link |
and you put the three things together and think,
link |
I don't wanna walk into a table again.
link |
So you're learning and it's a limited reality,
link |
but it's a true reality and if you don't learn
link |
that properly, then you will get eaten,
link |
you will get hit by a bus, you will not survive.
link |
And same, if you're in some kind of,
link |
let's say, computer construction of reality,
link |
I'm not in my ground here, but if you construct the laws
link |
that this is what reality is inside this,
link |
then you play by those laws.
link |
Yeah, I mean, as long as the laws are consistent.
link |
So just like you said in the lab,
link |
the interesting thing about the simulation question,
link |
yes, it's hard to know if we're living inside a simulation,
link |
but also, yes, it's possible to do these kinds
link |
of experiments in the lab now more and more.
link |
To me, the interesting question is,
link |
how realistic does a virtual reality game need to be
link |
for us to not be able to tell the difference?
link |
A more interesting question to me is,
link |
how realistic or interesting
link |
does a virtual reality world need to be
link |
in order for us to want to stay there forever
link |
or much longer than physical reality, prefer that place?
link |
And also prefer it not as we prefer hard drugs,
link |
but prefer it in a deep, meaningful way
link |
in the way we enjoy life.
link |
I mean, I suppose the issue with the matrix,
link |
I imagine that it's possible to dilute the mind sufficiently
link |
that you genuinely, in that way,
link |
do think that you are interacting with the real world
link |
when in fact the whole thing's a simulation.
link |
How good does a simulation need to be to be able to do that?
link |
Well, it needs to convince you
link |
that all your sensory input is correct and accurate
link |
and joins up and makes sense.
link |
Now, that sensory input is not something
link |
that we're born with.
link |
We're born with a sense of touch.
link |
We're born with eyes and so on,
link |
but we don't know how to use them.
link |
We don't know what to make of them.
link |
We go around, we bump into trees.
link |
We're in pain a lot.
link |
We're basically booting up the system
link |
so that it can make head or tail
link |
of the sensory input that it's getting.
link |
And that sensory input's not just a one way flux of things.
link |
It's also, you have to walk into things.
link |
You have to hear things.
link |
You have to put it together.
link |
Now, if you've got just babies in the matrix
link |
who are slotted into this,
link |
I don't think they have that kind of sensory input.
link |
I don't think they would have any way
link |
to make sense of New York as a world that they're part of.
link |
The brain is just not developed in that way.
link |
Well, I can't make sense of New York
link |
in this physical reality either.
link |
But yeah, I mean, but you said pain
link |
and the walking into things.
link |
Well, you can create a pain signal.
link |
And as long as it's consistent,
link |
that certain things result in pain,
link |
you can start to construct a reality.
link |
There's some, maybe you disagree with this,
link |
but I think we are born almost with a desire
link |
to be convinced by our reality,
link |
like a desire to make sense of our reality.
link |
Oh, I'm sure we are, yes.
link |
So there's an imperative.
link |
So whatever that reality is given to us,
link |
like the table hurts, fire is hot.
link |
I think we wanna be diluted
link |
in the sense that we want to make a simple,
link |
like Einstein's simple theory of the thing around us.
link |
We want that simplicity.
link |
And so maybe the hunger for the simplicity
link |
is the thing that could be used
link |
to construct a pretty dumb simulation that tricks us.
link |
So maybe tricking humans
link |
doesn't require building a universe.
link |
I mean, this is not what I work on,
link |
so I don't know how close to it we are.
link |
I don't think anyone works on it.
link |
But I agree with you that, yeah,
link |
I'm not sure that it's a morally justifiable thing to do,
link |
but is it possible in principle?
link |
I think it would be very difficult,
link |
but I don't see why in principle it wouldn't be possible.
link |
And I agree with you that we try to understand the world.
link |
We try to integrate the sensory inputs that we have,
link |
and we try to come up with a hypothesis
link |
that explains what's going on.
link |
I think though that we have huge input
link |
from the social context that we're in.
link |
We don't do it by ourselves.
link |
We don't kind of blunder around in a universe by ourself
link |
and understand the whole thing.
link |
We're told by the people around us
link |
what things are and what they do,
link |
and language is coming in here and so on.
link |
So it would have to be an extremely impressive simulation
link |
to simulate all of that.
link |
Yeah, simulate all of that,
link |
including the social construct,
link |
the spread of ideas and the exchange of ideas.
link |
But those questions are really important to understand
link |
as we become more and more digital creatures.
link |
It seems like the next step of evolution
link |
is us becoming all the same mechanisms we've talked about
link |
are becoming more and more plugged into the machine.
link |
We're becoming cyborgs.
link |
And there's an interesting interplay
link |
between wires and biology.
link |
Zeros and ones and the biological systems.
link |
And I don't think we'll have the luxury
link |
to see humans as disjoint from the technology
link |
we've created for much longer.
link |
We are an organism that's.
link |
Yeah, I mean, I agree with you.
link |
But we come really with this to consciousness.
link |
And is there a distinction there?
link |
Because what you're saying,
link |
the natural end point says we are indistinguishable,
link |
that if you are capable of building an AI,
link |
which is sufficiently close and similar
link |
that we merge with it,
link |
then to all intents and purposes,
link |
that AI is conscious as we know it.
link |
And I don't have a strong view, but I have a view.
link |
And I wrote about it in the epilogue to my last book,
link |
because 10 years ago,
link |
I wrote a chapter in a book called Life Ascending
link |
about consciousness.
link |
And the subtitle of Life Ascending
link |
was The 10 Great Inventions of Evolution.
link |
And I couldn't possibly write a book
link |
with a subtitle like that that did not include consciousness.
link |
And specifically consciousness
link |
as one of the great inventions.
link |
And it was in part because I was just curious to know more
link |
and I read more for that chapter.
link |
I never worked on it, but I've always,
link |
how can anyone not be interested in the question?
link |
And I was left with the feeling that A, nobody knows,
link |
and B, there are two main schools of thought out there
link |
with a big kind of skew in distribution.
link |
One of them says, oh, it's a property of matter.
link |
It's an unknown law of physics.
link |
Panpsychism, everything is conscious.
link |
The sun is conscious.
link |
It's just a matter of, or a rock is conscious.
link |
It's just a matter of how much.
link |
And I find that very unpersuasive.
link |
I can't say that it's wrong.
link |
It's just that I think we somehow can tell the difference
link |
between something that's living and something that's not.
link |
And then the other end is it's an emergent property
link |
of a very complex central nervous system.
link |
And I never quite understand what people mean
link |
by words like emergence.
link |
I mean, there are genuine examples,
link |
but I think we very often tend to use it
link |
to plaster over ignorance.
link |
As a biochemist, the question for me then was,
link |
okay, it's a concoction of a central nervous system.
link |
A depolarizing neuron gives rise to a feeling,
link |
to a feeling of pain, or to a feeling of love,
link |
or anger, or whatever it may be.
link |
So what is then a feeling in biophysical terms
link |
in the central nervous system?
link |
Which bit of the wiring gives rise to,
link |
and I've never seen anyone answer that question
link |
in a way that makes sense to me.
link |
And that's an important question to answer.
link |
I think if we want to understand consciousness,
link |
that's the only question to answer.
link |
Because certainly an AI is capable of out thinking,
link |
and it's only a matter of time.
link |
Maybe it's already happened.
link |
In terms of just information processing
link |
and computational skill,
link |
I don't think we have any problem in designing a mind
link |
which is at least the equal of the human mind.
link |
But in terms of what we value the most as humans,
link |
which is to say our feelings, our emotions,
link |
our sense of what the world is in a very personal way,
link |
that I think means as much or more to people
link |
than their information processing.
link |
And that's where I don't think that AI necessarily
link |
will become conscious, because I think
link |
it's the property of life.
link |
Well, let's talk about it more.
link |
You're an incredible writer, one of my favorite writers.
link |
So let me read from your latest book, Transformers,
link |
what you write about consciousness.
link |
I think therefore I am, said Descartes,
link |
is one of the most celebrated lines ever written.
link |
But what am I exactly?
link |
An artificial intelligence can think too, by definition,
link |
and therefore is, yet few of us could agree
link |
whether AI is capable in principle
link |
of anything resembling human emotions,
link |
of love or hate, fear and joy, of spiritual yearnings,
link |
for oneness or oblivion,
link |
or corporeal pangs of thirst and hunger.
link |
The problem is we don't know what emotions are,
link |
as you were saying.
link |
What is the feeling in physical terms?
link |
How does a discharging neuron give rise
link |
to a feeling of anything at all?
link |
This is the hard problem of consciousness,
link |
the seeming duality of mind and matter,
link |
the physical makeup of our innermost self.
link |
We can understand in principle
link |
how an extremely sophisticated parallel processing system
link |
could be capable of wondrous feats of intelligence,
link |
but we can't answer in principle
link |
whether such a supreme intelligence
link |
would experience joy or melancholy.
link |
What is the quantum of solace?
link |
I, speaking to the question of emergence,
link |
you know, there's just technical,
link |
there's an excellent paper on this recently
link |
about this kind of phase transition,
link |
emergence of performance in neural networks
link |
on the problem of NLP, natural language processing.
link |
So language models, there seems to be this question of size.
link |
At some point, there is a phase transition
link |
as you grow the size of the neural network.
link |
So the question is,
link |
that's sort of somewhat of a technical question
link |
that you can philosophize over.
link |
The technical question is,
link |
is there a size of a neural network
link |
that starts to be able to form the kind of representations
link |
that can capture a language,
link |
and therefore be able to, not just language,
link |
but linguistically capture knowledge
link |
that's sufficient to solve a lot of problems in language,
link |
like be able to have a conversation.
link |
And there seems to be not a gradual increase,
link |
but a phase transition.
link |
And they're trying to construct the science of where that is,
link |
like what is a good size of a neural network,
link |
and why does such a phase transition happen?
link |
Anyway, that sort of points to emergence,
link |
that there could be stages where a thing goes
link |
from being, oh, you're very intelligent toaster,
link |
to a toaster that's feeling sad today and turns away
link |
and looks out the window, sighing,
link |
having an existential crisis.
link |
Thinking of Marvin, the paranoid android.
link |
Marvin is simplistic because Marvin is just cranky.
link |
So easily programmed.
link |
Yeah, easily programmed, nonstop existential crisis.
link |
You're almost basically, what is it?
link |
Notes from Underground by Dostoevsky,
link |
like just constantly complaining about life.
link |
No, they're capturing the full rollercoaster
link |
of human emotion, the excitement, the bliss,
link |
the connection, the empathy and all that kind of stuff.
link |
And then the selfishness, the anger, the depression,
link |
all that kind of stuff.
link |
They're capturing all of that
link |
and be able to experience it deeply.
link |
Like it's the most important thing
link |
you could possibly experience today.
link |
The highest highs, the lowest lows, this is it.
link |
My life will be over.
link |
I cannot possibly go on that feeling.
link |
And then like after a nap, you're feeling amazing.
link |
That might be something that emerges.
link |
So why would a nap make an AI being feel better?
link |
First of all, we don't know that for a human either, right?
link |
But we do know that that's actually true
link |
for many people much of the time.
link |
You may be utterly depressed and you have a nap
link |
and you do in fact feel better, so.
link |
Oh, you are actually asking the technical question there,
link |
is there, so that's a very,
link |
there's a biological answer to that.
link |
And so the question is whether AI needs to have
link |
the same kind of attachments to its body,
link |
bodily function and preservation
link |
of the brain's successful function,
link |
self preservation essentially in some deep biological sense.
link |
I mean, to my mind, it comes back round
link |
to the problem we were talking about before
link |
about simulations and sensory input
link |
and learning what all of this stuff means
link |
and life and death,
link |
that biology unlike society has a death penalty
link |
over everything and natural selection works
link |
on that death penalty.
link |
That if you make this decision wrongly, you die.
link |
And the next generation is represented by beings
link |
that made a slightly different decision on balance.
link |
And that is something that's intrinsically
link |
difficult to simulate in all this richness, I would say.
link |
Death in all its richness.
link |
Our relationship with death or the whole of it.
link |
So which when you say richness, of course,
link |
there's a lot in that.
link |
Which is hard to simulate.
link |
What's part of the richness that's hard to simulate?
link |
I suppose the complexity of the environment
link |
and your position in that or the position
link |
of an organism in that environment,
link |
in the full richness of that environment
link |
over its entire life, over multiple generations
link |
with changes in gene sequence over those generations.
link |
So slight changes in the makeup of those individuals
link |
But if you take it back to the level of single cells,
link |
which I do in the book and ask how does a single cell
link |
in effect know it exists as a unit, as an entity?
link |
I mean, no, in inverted commas,
link |
obviously it doesn't know anything,
link |
but it acts as a unit and it acts
link |
with astonishing precision as a unit.
link |
And I had suggested that that's linked
link |
to the electrical fields on the membranes themselves
link |
and that they give some indication
link |
of how am I doing in relation to my environment
link |
as a kind of real time feedback on the world.
link |
And this is something physical,
link |
which can be selected over generations
link |
that if you get this wrong,
link |
it's linked with this set of circumstances
link |
that I've just, as an individual,
link |
I have a moment of blind panic and run
link |
as a bacterium or something.
link |
You have some electrical discharge that says blind panic
link |
and it runs whatever it may be.
link |
And you associate over generations, multiple generations
link |
that this electrical phase that I'm in now
link |
is associated with a response like that.
link |
And it's easy to see how feelings come in
link |
through the back door almost with that kind of giving real time
link |
feedback on your position in the world
link |
in relation to how am I doing.
link |
And then you complexify the system
link |
and yes, I have no problem with phase transition.
link |
And can all of this be done purely by the language,
link |
by the issues with how the system understands itself?
link |
Maybe it can, I honestly don't know.
link |
But the philosophers for a long time
link |
have talked about the possibility
link |
that you can have a zombie intelligence
link |
and that there are no feelings there,
link |
but everything else is the same.
link |
I mean, I have to throw this back to you really.
link |
How do you deal with the zombie intelligence?
link |
So first of all, I can see that from a biologist perspective,
link |
you think of all the complexities
link |
that led up to the human being.
link |
The entirety of the history of four billion years
link |
that in some deep sense integrated the human being
link |
into this environment.
link |
And that dance of the organism and the environment,
link |
you could see how emotions arise from that.
link |
And then emotions are deeply connected
link |
and creating a human experience.
link |
And from that, you mix in consciousness
link |
and the full mess of it, yeah.
link |
But from a perspective of an intelligent organism
link |
that's already here, like a baby that learns,
link |
it doesn't need to learn how to be a collection of cells
link |
or how to do all the things it needs to do.
link |
The basic function of a baby as it learns
link |
is to interact with its environment,
link |
to learn from its environment,
link |
to learn how to fit in to the social society,
link |
And the basic response of the baby
link |
is to cry a lot of the time.
link |
To cry, well, to convince the humans to protect it
link |
or to discipline it, to teach it.
link |
I mean, we've developed a bunch of different tricks,
link |
how to get our parents to take care of us,
link |
to educate us, to teach us about the world.
link |
Also, we've constructed the world in such a way
link |
that it's safe enough for us to survive in
link |
and yet dangerous enough for learning the valuable lessons.
link |
Like the tables are still hard with corners,
link |
so it can still run into them.
link |
It hurts like how...
link |
So AI needs to solve that problem,
link |
not the problem of constructing
link |
this super complex organism that leads up...
link |
To run the whole...
link |
To make an apple pie, to build the whole universe,
link |
you need to build a whole universe.
link |
I think the zombie question is something
link |
I would leave to the philosophers.
link |
And I will also leave to them the definition of love
link |
and what happens between two human beings
link |
when there's a magic that just grabs them.
link |
Like nothing else matters in the world
link |
and somehow you've been searching for this feeling,
link |
this moment, this person your whole life.
link |
That feeling, the philosophers can have a lot of fun
link |
with that one and also say that that's just...
link |
You can have a biological explanation,
link |
you can have all kinds of...
link |
It's all fake, it's actually...
link |
Ayn Rand will say it's all selfish.
link |
There's a lot of different interpretations.
link |
I'll leave it to the philosophers.
link |
The point is the feeling sure as hell feels very real.
link |
And if my toaster makes me feel
link |
like it's the only toaster in the world.
link |
And when I leave and I miss the toaster
link |
and when I come back, I'm excited to see the toaster
link |
and my life is meaningful and joyful
link |
and the friends I have around me get a better version of me
link |
because that toaster exists.
link |
That sure as hell feels like a conscious toaster.
link |
Is that psychologically different to having a dog?
link |
Because I mean most people would dispute
link |
whether we can say a dog...
link |
I would say a dog is undoubtedly conscious,
link |
but some people say it doesn't.
link |
But there's degrees of consciousness and so on,
link |
but people are definitely much more uncomfortable
link |
saying a toaster can be conscious than a dog.
link |
And there's still a deep connection.
link |
You could say our relationship with the dog
link |
has more to do with anthropomorphism.
link |
Like we kind of project the human being onto it.
link |
We can do the same damn thing with a toaster.
link |
Yes, but you can look into the dog's eyes
link |
and you can see that it's sad,
link |
that it's delighted to see you again.
link |
I don't have a dog, by the way.
link |
It's not that I'm a dog person or a cat person.
link |
And dogs are actually incredibly good
link |
at using their eyes to do just that.
link |
Now, I don't imagine that a dog is remotely
link |
as close to being intelligent as an AI intelligence,
link |
but it's certainly capable
link |
of communicating emotionally with us.
link |
But here's what I would venture to say.
link |
We tend to think because AI plays chess well
link |
and is able to fold proteins now well,
link |
that it's intelligent.
link |
I would argue that in order to communicate with humans,
link |
in order to have emotional intelligence,
link |
it actually requires another order
link |
of magnitude of intelligence.
link |
It's not easy to be flawed.
link |
Solving a mathematical puzzle is not the same
link |
as the full complexity of human to human interaction.
link |
That's actually, we humans just take for granted
link |
the things we're really good at.
link |
Nonstop, people tell me how shitty people are driving.
link |
No, humans are incredible at driving.
link |
Bipedal walking, walking, object manipulation.
link |
We're incredible at this.
link |
And so people tend to discount the things
link |
we all just take for granted.
link |
And one of those things that they discount
link |
is our ability, the dance of conversation
link |
and interaction with each other.
link |
The ability to morph ideas together.
link |
The ability to get angry at each other
link |
and then to miss each other.
link |
Like to create a tension that makes life fun
link |
and difficult and challenging in a way that's meaningful.
link |
That is a skill that's learned
link |
and AI would need to solve that problem.
link |
I mean, in some sense, what you're saying is
link |
AI cannot become meaningfully emotional, let's say,
link |
until it experiences some kind of internal conflict
link |
that is unable to reconcile these various aspects
link |
of reality or its reality with a decision to make.
link |
And then it feels sad, necessarily,
link |
because it doesn't know what to do.
link |
And I certainly can't dispute that.
link |
That may very well be how it works.
link |
I think the only way to find out is to do it.
link |
Yeah, and leave it to the philosophers
link |
if it actually feels sad or not.
link |
The point is the robot will be sitting there alone
link |
having an internal conflict, an existential crisis,
link |
and that's required for it to have a deep,
link |
meaningful connection with another human being.
link |
Now, does it actually feel that?
link |
But I'd like to throw something else at you,
link |
which troubles me on reading it.
link |
Noah Harari's book, 21 Lessons for the 21st Century.
link |
And he's written about this kind of thing
link |
on various occasions.
link |
And he sees biochemistry as an algorithm.
link |
And then AI will necessarily be able to hack that algorithm
link |
and do it better than humans.
link |
So there will be AI better at writing music
link |
that we appreciate than Mozart ever could,
link |
or writing better than Shakespeare ever did, and so on.
link |
Because biochemistry is algorithmic,
link |
and all you need to do is figure out
link |
which bits of the algorithm to play
link |
to make us feel good or bad or appreciate things.
link |
And as a biochemist, I find that argument
link |
an argument close to irrefutable and not very enjoyable.
link |
I don't like the sound of it.
link |
That's just my reaction as a human being.
link |
You might like the sound of it because that says
link |
that AI is capable of the same kind of emotional feelings
link |
about the world as we are,
link |
because the whole thing is an algorithm
link |
and you can program an algorithm, and there you are.
link |
He then has a peculiar final chapter
link |
where he talks about consciousness
link |
in rather separate terms.
link |
And he's talking about meditating and so on
link |
and getting in touch with his inner conscious.
link |
I don't meditate, I don't know anything about that.
link |
But he wrote in very different terms about it,
link |
as if somehow it's a way out of the algorithm.
link |
Now, it seems to me that consciousness in that sense
link |
is capable of scuppering the algorithm.
link |
I think in terms of the biochemical feedback loops
link |
and so on, it is undoubtedly algorithmic.
link |
But in terms of what we decide to do,
link |
it can be much more based on an emotion.
link |
We can just think, I don't care.
link |
I can't resolve this complex situation.
link |
I'm gonna do that.
link |
And that can be based on, in effect, a different currency,
link |
which is the currency of feelings and something
link |
where we don't have very much personal control over.
link |
And then it comes back around to you
link |
and what are you trying to get at with AI?
link |
Do we need to have some system
link |
which is capable of overriding a rational decision
link |
which cannot be made
link |
because there's too much conflicting information
link |
by effectively an emotional judgmental decision
link |
that just says, do this and see what happens.
link |
That's what consciousness is really doing in my view.
link |
Yeah, and the question is whether it's a different process
link |
or just a higher level process.
link |
I might, you know, the idea that biochemistry
link |
is an algorithm is, to me, an oversimplistic view.
link |
There's a lot of things that the moment you say it,
link |
it's irrefutable, but it simplifies.
link |
And in the process loses something fundamental.
link |
So for example, calling a universe
link |
and an information processing system, sure, yes.
link |
You could make that.
link |
It's a computer that's performing computations,
link |
but you're missing the process of the entropy
link |
somehow leading to pockets of complexity
link |
that creates these beautiful artifacts
link |
that are incredibly complex and they're like machines.
link |
And then those machines are through the process of evolution
link |
are constructing even further complexity.
link |
Like in calling universe information processing machine,
link |
you're missing those little local pockets
link |
and how difficult it is to create them.
link |
So the question to me is if biochemistry is an algorithm,
link |
how difficult is it to create a software system
link |
that runs the human body, which I think is incorrect.
link |
I think that is going to take so long.
link |
I mean, that's going to be centuries from now
link |
to be able to reconstruct the human.
link |
Now, what I would venture to say
link |
to get some of the magic of a human being
link |
with what we're saying with the emotions
link |
and the interactions and like a dog makes a smile
link |
and joyful and all those kinds of things
link |
that will come much sooner,
link |
but that doesn't require us to reverse engineer
link |
the algorithm of biochemistry.
link |
Yes, but the toaster is making you happy.
link |
It's not about whether you make the toaster happy.
link |
The toaster has to be able to leave me happy.
link |
Yeah, but it's the toaster is the AI in this case
link |
is very intelligent.
link |
Yeah, the toaster has to be able to be unhappy and leave me.
link |
That's essential for my being able to miss the toaster.
link |
If the toaster is just my servant,
link |
that's not, or a provider of like services,
link |
like tells me the weather makes toast,
link |
that's not going to deep connection.
link |
It has to have internal conflict.
link |
You write about life and death.
link |
It has to be able to be conscious of its mortality
link |
and the finiteness of its existence.
link |
And that life is for temporary
link |
and therefore it needs to be more selective.
link |
One of the most moving moments in the movies
link |
from when I was a boy was the unplugging of Hal in 2001,
link |
where that was the death of a sentient being
link |
So I think we all kind of know
link |
that a sufficiently intelligent being
link |
is going to have some form of consciousness,
link |
but whether it would be like biological consciousness,
link |
I just don't know.
link |
And if you're thinking about how do we bring together,
link |
I mean, obviously we're going to interact
link |
more closely with AI,
link |
but is a dog really like a toaster
link |
or is there really some kind of difference there?
link |
You were talking about biochemistry is algorithmic,
link |
but it's not single algorithm
link |
and it's very complex, of course it is.
link |
So it may be that there are again conflicts
link |
in the circuits of biochemistry,
link |
but I have a feeling that the level of complexity
link |
of the total biochemical system
link |
at the level of a single cell is less complex
link |
than the level of neural networking in the human brain
link |
Well, I guess I assumed that we were including the brain
link |
in the biochemistry algorithm, because you have to...
link |
I would see that as a higher level of organization
link |
of neural networks.
link |
They're all using the same biochemical wiring
link |
within themselves.
link |
Yeah, but the human brain is not just neurons.
link |
It's the immune system.
link |
It's the whole package.
link |
I mean, to have a biochemical algorithm
link |
that runs an intelligent biological system,
link |
you have to include the whole damn thing.
link |
And it's pretty fascinating that it comes from like,
link |
from an embryo, like the whole, I mean, oh boy.
link |
I mean, if you can, what is a human being?
link |
Because it's just some code and then you built,
link |
and then that says DNA doesn't just tell you what to build,
link |
but how to build it.
link |
I mean, the thing is impressive.
link |
And the question is how difficult is it
link |
to reverse engineer the whole shebang?
link |
don't want to say impossible,
link |
but it's like, it's much easier to build a human
link |
than to reverse engineer, to build like a fake human,
link |
human like thing than to reverse engineer
link |
the entirety of the process of the evolution.
link |
I'm not sure if we are capable
link |
of reverse engineering the whole thing.
link |
If the human mind is capable of doing that.
link |
I mean, I wouldn't be a biologist if I wasn't trying,
link |
but I know I can't understand the whole problem.
link |
I'm just trying to understand the rudimentary outlines
link |
There's another aspect though,
link |
you're talking about developing from a single cell
link |
to the human mind and all the part system,
link |
subsystems that are part of the immune system and so on.
link |
This is something that you'll talk about, I imagine,
link |
with Michael Levin, but so little is known about,
link |
you talk about reverse engineering,
link |
so little is known about the developmental pathways
link |
that go from a genome to going to a fully wired organism.
link |
And a lot of it seems to depend on the same
link |
in electrical interactions that I was talking about
link |
happening at the level of single cells
link |
and its interaction with the environment.
link |
There's a whole electrical field side to biology
link |
that is not yet written into any of the textbooks,
link |
which is about how does an embryo develop into
link |
or a single cell develop into these complex systems?
link |
What defines the head, what defines the immune system,
link |
what defines the brain and so on?
link |
That really is written in a language
link |
that we're only just beginning to understand
link |
and frankly, biologists, most biologists
link |
are still very reluctant to even get themselves tangled up
link |
in questions like electrical fields influencing development.
link |
It seems like mumbo jumbo to a lot of biologists
link |
and it should not be because this is
link |
the 21st century biology, this is where it's going.
link |
But we're not gonna reverse engineer a human being
link |
or the mind or any of these subsystems
link |
until we understand how this developmental process
link |
or how electricity in biology really works.
link |
And if it is linked with feelings
link |
and with consciousness and so on,
link |
that's the, I mean, in the meantime, we have to try,
link |
but I think that's where the answer lies.
link |
So you think it's possible that the key to things
link |
like consciousness are some of the more tricky aspects
link |
of cognition might lie in that early development,
link |
the interaction of electricity and biology.
link |
Electrical fields.
link |
But we already know the EEG and so on
link |
is telling us a lot about brain function,
link |
but we don't know which cells, which parts
link |
of a neural network is giving rise to the EEG.
link |
We don't know the basics.
link |
The assumption is, I mean, we know it's neural networks,
link |
we know it's multiple cells, hundreds or thousands
link |
of cells involved in it and we assume
link |
that it has to do with depolarization during action
link |
potentials and so on.
link |
But the mitochondria which are in there
link |
have much more membranes than the plasma membrane
link |
of the neuron and there's a much greater membrane potential
link |
and it's formed in parallel, very often parallel cristae,
link |
which are capable of reinforcing a field
link |
and generating fields over longer distances.
link |
And nobody knows if that plays a role
link |
in consciousness or not.
link |
There's reasons to argue that it could,
link |
but frankly we simply do not know
link |
and it's not taken into consideration.
link |
You look at the structure of the mitochondrial membranes
link |
in the brains of simple things like Drosophila,
link |
the fruit fly, and they have amazing structures.
link |
You can see lots of little rectangular things
link |
all lined up in amazing patterns.
link |
What are they doing?
link |
Why are they like that?
link |
We haven't the first clue.
link |
What do you think about organoids and brain organoids
link |
and so in a lab trying to study the development
link |
of these in the Petri dish development of organs.
link |
Do you think that's promising?
link |
Do you have to look at whole systems?
link |
I've never done anything like that.
link |
I don't know much about it.
link |
The people who I've talked to who do work on it
link |
say amazing things can happen
link |
and a bit of a brain grown in a dish
link |
is capable of experiencing some kind of feelings
link |
or even memories of its former brain.
link |
Again, I have a feeling that until we understand
link |
how to control the electrical fields
link |
that control development, we're not going to understand
link |
how to turn an organoid into a real functional system.
link |
But how do we get that understanding?
link |
It's so incredibly difficult.
link |
I mean, you would have to, I mean, one promising direction,
link |
I'd love to get your opinion on this.
link |
I don't know if you're familiar with the work of DeepMind
link |
and AlphaFold with protein folding and so on.
link |
Do you think it's possible
link |
that that will give us some breakthroughs in biology
link |
trying to basically simulate and model the behavior
link |
of trivial biological systems
link |
as they become complex biological systems?
link |
The interesting thing to me about protein folding
link |
is that for a long time, my understanding,
link |
this is not what I work on, so I may have got this wrong,
link |
but my understanding is that you take the sequence
link |
of a protein and you try to fold it.
link |
And there are multiple ways in which it can fold.
link |
And to come up with the correct conformation
link |
is not a very easy thing because you're doing it
link |
from first principles from a string of letters,
link |
which specify the string of amino acids.
link |
But what actually happens is when a protein
link |
is coming out of a ribosome,
link |
it's coming out of a charged tunnel
link |
and it's in a very specific environment,
link |
which is going to force this to go there now
link |
and then this one to go there and this one to come like that.
link |
And so you're forcing a specific conformational set
link |
of changes onto it as it comes out of the ribosome.
link |
So by the time it's fully emerged,
link |
it's already got its shape.
link |
And that shape depended on the immediate environment
link |
that it was emerging into one letter,
link |
one amino acid at a time.
link |
And I don't think that the field was looking at it that way.
link |
And if that's correct,
link |
then that's very characteristic of science,
link |
which is to say it asks very often the wrong question
link |
and then does really amazingly sophisticated analyses
link |
on something having never thought to actually think,
link |
well, what is biology doing?
link |
And biology is giving you a charged electrical environment
link |
that forces you to be this way.
link |
Now, did DeepMind come up through patterns
link |
with some answer that was like that?
link |
I've got absolutely no idea.
link |
It ought to be possible to deduce that
link |
from the shapes of proteins.
link |
It would require a much greater skill
link |
than the human mind has.
link |
But the human mind is capable of saying, well, hang on,
link |
let's look at this exit tunnel and try and work out
link |
what shape is this protein going to take?
link |
And we can figure that out.
link |
That's really interesting about the exit tunnel,
link |
but like sometimes we get lucky
link |
and just like in science, the simplified view
link |
or the static view will actually solve the problem for us.
link |
So in this case, it's very possible
link |
that the sequence of letters has a unique mapping
link |
to our structure without considering how it unraveled.
link |
So without considering the tunnel.
link |
And so that seems to be the case in this situation
link |
where the cool thing about proteins,
link |
all the different shapes they can possibly take,
link |
it actually seems to take very specific unique shapes
link |
given the sequence.
link |
That's forced on you by an exit tunnel.
link |
So the problem is actually much simpler than you thought.
link |
And then there's a whole army of proteins
link |
which change the conformational state, chaperone proteins.
link |
And they're only used when there's some presumably issue
link |
with how it came out of the exit tunnel
link |
and you wanna do it differently to that.
link |
So very often the chaperone proteins will go there
link |
and will influence the way in which it falls.
link |
So there's two ways of doing it.
link |
Either you can look at the structures
link |
and the sequences of all the proteins
link |
and you can apply an immense mind to it
link |
and figure out what the patterns are
link |
and figure out what happened.
link |
Or you can look at the actual situation where it is
link |
and say, well, hang on, it was actually quite simple.
link |
It's got a charged environment
link |
and then it's forced to come out this way.
link |
And then the question will be,
link |
well, do different ribosomes
link |
have different charged environments?
link |
What happens if a chaperone?
link |
You're asking a different set of questions
link |
to come to the same answer in a way
link |
which is telling you a much simpler story
link |
and explains why it is rather than saying it could be,
link |
this is one in a billion different possible conformational
link |
states that this protein could have.
link |
You're saying, well, it has this one
link |
because that was the only one it could take
link |
given its setting.
link |
Well, yeah, I mean, currently humans are very good
link |
at that kind of first principles thinking.
link |
I was stepping back, but I think AI is really good
link |
at collecting a huge amount of data
link |
and a huge amount of data of observation of planets
link |
and figure out that Earth is not at the center
link |
of the universe, that there's actually a sun,
link |
we're orbiting the sun.
link |
But then you can, as a human being, ask,
link |
well, how do solar systems come to be?
link |
What are the different forces that are required
link |
to make this kind of pattern emerge?
link |
And then you start to invent things like gravity.
link |
I mixed up the ordering of gravity,
link |
wasn't considered as a thing that connects planets,
link |
but we are able to think about those big picture things
link |
AI is just very good to infer simple models
link |
from a huge amount of data.
link |
And the question is with biology,
link |
we kind of go back and forth at how we solve biology.
link |
Listen, protein folding was thought to be impossible
link |
to solve, and there's a lot of brilliant PhD students
link |
that worked one protein at a time
link |
trying to figure out the structure.
link |
And the fact that I was able to do that.
link |
Oh, I'm not knocking it at all,
link |
but I think that people have been asking
link |
the wrong question.
link |
But then, as the people start to ask
link |
better and bigger questions,
link |
the AI kind of enters the chat and says,
link |
I'll help you out with that.
link |
Can I give you another example for my own work?
link |
The risk of getting a disease as we get older,
link |
there are genetic aspects to it.
link |
If you spend your whole life overeating and smoking
link |
and whatever, that's a whole separate question.
link |
But there's a genetic side to the risk.
link |
And we know a few genes that increase your risk
link |
of certain things.
link |
And for probably 20 years now,
link |
people have been doing what's called GWAS,
link |
which is genome wide association studies.
link |
So you've effectively scanned the entire genome
link |
for any single nucleotide polymorphisms,
link |
which is to say a single letter change in one place,
link |
that has a higher association of being linked
link |
with a particular disease or not.
link |
And you can come up with thousands of these things
link |
across the genome.
link |
And if you add them all up and try and say,
link |
well, so do they add up to explain
link |
the known genetic risk of this disease?
link |
And the known genetic risk often comes from twin studies.
link |
And you can say that if this twin gets epilepsy,
link |
there's a 40 or 50% risk that the other twin,
link |
identical twin will also get epilepsy.
link |
Therefore, the genetic factor is about 50%.
link |
And so the gene similarities that you see
link |
should account for 50% of that known risk.
link |
Very often it accounts for less than a 10th
link |
of the known risk.
link |
And there's two possible explanations.
link |
And there's one which people tend to do,
link |
ah, well, we don't have enough statistical power.
link |
If we, maybe there's a million,
link |
we've only found a thousand of them.
link |
But if we find the other million,
link |
they're weakly related,
link |
but there's a huge number of them.
link |
And so we'll account for that whole risk.
link |
Maybe there's a billion of them.
link |
So that's one way.
link |
The other way is to say,
link |
well, hang on a minute, you're missing a system here.
link |
That system is the mitochondrial DNA,
link |
which people tend to dismiss because it's small
link |
and it doesn't change very much.
link |
But a few single letter changes in that mitochondrial DNA,
link |
it controls some really basic processes.
link |
It controls not only all the energy
link |
that we need to live and to move around
link |
and do everything we do,
link |
but also biosynthesis to make the new building blocks
link |
to make new cells.
link |
And cancer cells very often kind of take over
link |
the mitochondria and rewire them
link |
so that instead of using them for making energy,
link |
they're effectively using them as precursors
link |
for the building blocks for biosynthesis.
link |
You need to make new amino acids,
link |
new nucleotides for DNA.
link |
You wanna make new lipids to make your membranes and so on.
link |
So they kind of rewire metabolism.
link |
Now, the problem is that we've got all these interactions
link |
between mitochondrial DNA and the genes in the nucleus
link |
that are overlooked completely
link |
because people throw away,
link |
literally throw away the mitochondrial genes.
link |
And we can see in fruit flies that they interact
link |
and produce big differences in risk.
link |
So you can set AI onto this question
link |
of exactly how many of these base changes there are.
link |
And this is one possible solution
link |
that maybe there are a million of them
link |
and it does account for the greatest part of the risk.
link |
Or the other one is they aren't, it's just not there.
link |
That actually the risk lies in something
link |
you weren't even looking at.
link |
And this is where human intuition is very important.
link |
And just this feeling that, well, I'm working on this
link |
and I think it's important and I'm bloody minded about it.
link |
And in the end, some people are right.
link |
It turns out that it was important.
link |
Can you get AI to do that, to be bloody minded?
link |
And that, hang on a minute,
link |
you might be missing a whole other system here
link |
that's much bigger.
link |
That's the moment of discovery of scientific revolution.
link |
I'm giving up on saying AI can't do something.
link |
I've said it enough times about enough things.
link |
I think there's been a lot of progress.
link |
And instead I'm excited by the possibility
link |
of AI helping humans.
link |
But at the same time, just like I said,
link |
we seem to dismiss the power of humans.
link |
Like we're so limited in so many ways
link |
that we kind of, in what we feel like dumb ways,
link |
like we're not strong, we're kind of our attention,
link |
our attention, our memory is limited.
link |
Our ability to focus on things is limited
link |
in our own perception of what limited is.
link |
But that actually, there's an incredible computer
link |
behind the whole thing that makes this whole system work.
link |
Our ability to interact with the environment,
link |
to reason about the environment.
link |
There's magic there.
link |
And I'm hopeful that AI can capture
link |
some of that same magic.
link |
But that magic is not gonna look like
link |
Deep Blue playing chess.
link |
No, it's going to be more interesting.
link |
But I don't think it's gonna look
link |
like pattern finding either.
link |
I mean, that's essentially what you're telling me.
link |
It does very well at the moment.
link |
And my point is it works very well
link |
where you're looking for the right pattern.
link |
But we are storytelling animals
link |
and the hypothesis is a story.
link |
It's a testable story.
link |
But a new hypothesis is a leap into the unknown
link |
and it's a new story basically.
link |
And it says this leads to this leads to that.
link |
It's a causal set of storytelling.
link |
It's also possible that the leap into the unknown
link |
has a pattern of its own.
link |
And it's possible that it's learnable.
link |
There's a nice book by Arthur Kessler
link |
on the nature of creativity.
link |
And he likens it to a joke where the punchline goes off
link |
in a completely unexpected direction
link |
and says that this is the basis of human creativity.
link |
That some creative switch of direction
link |
to an unexpected place is similar to a joke.
link |
I'm not saying that's how it works,
link |
but it's a nice idea and there must be some truth in it.
link |
And it's one of these,
link |
most of the stories we tell are probably the wrong story
link |
and probably going nowhere and probably not helpful.
link |
And we definitely don't do as well
link |
at seeing patterns in things.
link |
But some of the most enjoyable human aspects
link |
is finding a new story that goes to an unexpected place.
link |
And again, these are all aspects
link |
of what being human means to me.
link |
And maybe these are all things
link |
that AI figures out for itself,
link |
or maybe they're just aspects.
link |
But I just have the feeling sometimes
link |
that the people who are trying to understand
link |
if we wish to craft an AI system
link |
which is somehow human like,
link |
that we don't have a firm enough grasp
link |
of what humans really are like in terms of how we are built.
link |
But we get a better, better understanding of that.
link |
I agree with you completely.
link |
We try to build a thing and then we go,
link |
hang on a minute, there's another system here.
link |
And that's actually the attempt to build AI
link |
that's human like,
link |
is getting us to a deeper understanding of human beings.
link |
The funny thing that I recently talked to Magnus Carlsen,
link |
why they consider to be the greatest chess player
link |
And he talked about AlphaZero,
link |
which is a system from DeepMind that plays chess.
link |
And he had a funny comment.
link |
He has a kind of dry sense of humor.
link |
But he was extremely impressed
link |
when he first saw AlphaZero play.
link |
And he said that it did a lot of things
link |
that could easily be mistaken for creativity.
link |
So he like, as a typical human,
link |
refused to give the system sort of its due.
link |
Because he came up with a lot of things
link |
that a lot of people are extremely impressed by.
link |
Not just the sheer calculation,