back to indexRay Kurzweil: Singularity, Superintelligence, and Immortality | Lex Fridman Podcast #321
link |
By the time he gets to 2045,
link |
we'll be able to multiply our intelligence
link |
many millions fold.
link |
And it's just very hard to imagine what that will be like.
link |
The following is a conversation with Ray Kurzweil,
link |
author, inventor, and futurist,
link |
who has an optimistic view of our future
link |
as a human civilization,
link |
predicting that exponentially improving technologies
link |
will take us to a point of a singularity
link |
beyond which superintelligent artificial intelligence
link |
will transform our world in nearly unimaginable ways.
link |
18 years ago, in the book Singularity is Near,
link |
he predicted that the onset of the singularity
link |
will happen in the year 2045.
link |
He still holds to this prediction and estimate.
link |
In fact, he's working on a new book on this topic
link |
that will hopefully be out next year.
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 Ray Kurzweil.
link |
In your 2005 book titled The Singularity is Near,
link |
you predicted that the singularity will happen in 2045.
link |
So now, 18 years later, do you still estimate
link |
that the singularity will happen on 2045?
link |
And maybe first, what is the singularity,
link |
the technological singularity, and when will it happen?
link |
Singularity is where computers really change our view
link |
of what's important and change who we are.
link |
But we're getting close to some salient things
link |
that will change who we are.
link |
A key thing is 2029,
link |
when computers will pass the Turing test.
link |
And there's also some controversy
link |
whether the Turing test is valid.
link |
Most people do believe that,
link |
but there's some controversy about that.
link |
But Stanford got very alarmed at my prediction about 2029.
link |
I made this in 1999 in my book.
link |
The Age of Spiritual Machines.
link |
And then you repeated the prediction in 2005.
link |
So they held an international conference,
link |
you might have been aware of it,
link |
of AI experts in 1999 to assess this view.
link |
So people gave different predictions,
link |
and they took a poll.
link |
It was really the first time that AI experts worldwide
link |
were polled on this prediction.
link |
And the average poll was 100 years.
link |
20% believed it would never happen.
link |
And that was the view in 1999.
link |
80% believed it would happen,
link |
but not within their lifetimes.
link |
There's been so many advances in AI
link |
that the poll of AI experts has come down over the years.
link |
So a year ago, something called Meticulous,
link |
which you may be aware of,
link |
assesses different types of experts on the future.
link |
They again assessed what AI experts then felt.
link |
And they were saying 2042.
link |
For the Turing test.
link |
For the Turing test.
link |
So it's coming down.
link |
And I was still saying 2029.
link |
A few weeks ago, they again did another poll,
link |
So AI experts now basically agree with me.
link |
I haven't changed at all, I've stayed with 2029.
link |
And AI experts now agree with me,
link |
but they didn't agree at first.
link |
So Alan Turing formulated the Turing test,
link |
Right, now, what he said was very little about it.
link |
I mean, the 1950 paper
link |
where he had articulated the Turing test,
link |
there's like a few lines that talk about the Turing test.
link |
And it really wasn't very clear how to administer it.
link |
And he said if they did it in like 15 minutes,
link |
that would be sufficient,
link |
which I don't really think is the case.
link |
These large language models now,
link |
some people are convinced by it already.
link |
I mean, you can talk to it and have a conversation with it.
link |
You can actually talk to it for hours.
link |
So it requires a little more depth.
link |
There's some problems with large language models
link |
which we can talk about.
link |
But some people are convinced by the Turing test.
link |
Now, if somebody passes the Turing test,
link |
what are the implications of that?
link |
Does that mean that they're sentient,
link |
that they're conscious or not?
link |
It's not necessarily clear what the implications are.
link |
Anyway, I believe 2029, that's six, seven years from now,
link |
we'll have something that passes the Turing test
link |
and a valid Turing test,
link |
meaning it goes for hours, not just a few minutes.
link |
Can you speak to that a little bit?
link |
What is your formulation of the Turing test?
link |
You've proposed a very difficult version
link |
of the Turing test, so what does that look like?
link |
Basically, it's just to assess it over several hours
link |
and also have a human judge that's fairly sophisticated
link |
on what computers can do and can't do.
link |
If you take somebody who's not that sophisticated
link |
or even an average engineer,
link |
they may not really assess various aspects of it.
link |
So you really want the human to challenge the system.
link |
On its ability to do things
link |
like common sense reasoning, perhaps.
link |
That's actually a key problem with large language models.
link |
They don't do these kinds of tests
link |
that would involve assessing chains of reasoning,
link |
but you can lose track of that.
link |
If you talk to them,
link |
they actually can talk to you pretty well
link |
and you can be convinced by it,
link |
but it's somebody that would really convince you
link |
that it's a human, whatever that takes.
link |
Maybe it would take days or weeks,
link |
but it would really convince you that it's human.
link |
Large language models can appear that way.
link |
You can read conversations and they appear pretty good.
link |
There are some problems with it.
link |
It doesn't do math very well.
link |
You can ask how many legs did 10 elephants have
link |
and they'll tell you, well, okay,
link |
each elephant has four legs
link |
and it's 10 elephants, so it's 40 legs.
link |
And you go, okay, that's pretty good.
link |
How many legs do 11 elephants have?
link |
And they don't seem to understand the question.
link |
Do all humans understand that question?
link |
No, that's the key thing.
link |
I mean, how advanced a human do you want it to be?
link |
But we do expect a human
link |
to be able to do multi chain reasoning,
link |
to be able to take a few facts
link |
and put them together, not perfectly.
link |
And we see that in a lot of polls
link |
that people don't do that perfectly at all.
link |
So it's not very well defined,
link |
but it's something where it really would convince you
link |
that it's a human.
link |
Is your intuition that large language models
link |
will not be solely the kind of system
link |
that passes the Turing test in 2029?
link |
Do we need something else?
link |
No, I think it will be a large language model,
link |
but they have to go beyond what they're doing now.
link |
I think we're getting there.
link |
And another key issue is if somebody
link |
actually passes the Turing test validly,
link |
I would believe they're conscious.
link |
And then not everybody would say that.
link |
It's okay, we can pass the Turing test,
link |
but we don't really believe that it's conscious.
link |
That's a whole nother issue.
link |
But if it really passes the Turing test,
link |
I would believe that it's conscious.
link |
But I don't believe that of large language models today.
link |
If it appears to be conscious,
link |
that's as good as being conscious, at least for you,
link |
I mean, consciousness is not something that's scientific.
link |
I mean, I believe you're conscious,
link |
but it's really just a belief,
link |
and we believe that about other humans
link |
that at least appear to be conscious.
link |
When you go outside of shared human assumption,
link |
like are animals conscious?
link |
Some people believe they're not conscious.
link |
Some people believe they are conscious.
link |
And would a machine that acts just like a human be conscious?
link |
I mean, I believe it would be.
link |
But that's really a philosophical belief.
link |
You can't prove it.
link |
I can't take an entity and prove that it's conscious.
link |
There's nothing that you can do
link |
that would indicate that.
link |
It's like saying a piece of art is beautiful.
link |
Multiple people can experience a piece of art as beautiful,
link |
but you can't prove it.
link |
But it's also an extremely important issue.
link |
I mean, imagine if you had something
link |
where nobody's conscious.
link |
The world may as well not exist.
link |
And so some people, like say Marvin Minsky,
link |
said, well, consciousness is not logical,
link |
it's not scientific, and therefore we should dismiss it,
link |
and any talk about consciousness is just not to be believed.
link |
But when he actually engaged with somebody
link |
who was conscious, he actually acted
link |
as if they were conscious.
link |
He didn't ignore that.
link |
He acted as if consciousness does matter.
link |
Whereas he said it didn't matter.
link |
Well, that's Marvin Minsky.
link |
He's full of contradictions.
link |
But that's true of a lot of people as well.
link |
But to you, consciousness matters.
link |
But to me, it's very important.
link |
But I would say it's not a scientific issue.
link |
It's a philosophical issue.
link |
And people have different views.
link |
Some people believe that anything
link |
that makes a decision is conscious.
link |
So your light switch is conscious.
link |
Its level of consciousness is low,
link |
not very interesting, but that's a consciousness.
link |
So a computer that makes a more interesting decision
link |
is still not at human levels,
link |
but it's also conscious and at a higher level
link |
than your light switch.
link |
So that's one view.
link |
There's many different views of what consciousness is.
link |
So if a system passes the Turing test,
link |
it's not scientific, but in issues of philosophy,
link |
things like ethics start to enter the picture.
link |
Do you think there would be,
link |
we would start contending as a human species
link |
about the ethics of turning off such a machine?
link |
Yeah, I mean, that's definitely come up.
link |
Hasn't come up in reality yet.
link |
But I'm talking about 2029.
link |
It's not that many years from now.
link |
So what are our obligations to it?
link |
It has a different, I mean, a computer that's conscious,
link |
it has a little bit different connotations than a human.
link |
We have a continuous consciousness.
link |
We're in an entity that does not last forever.
link |
Now, actually, a significant portion of humans still exist
link |
and are therefore still conscious.
link |
But anybody who is over a certain age
link |
doesn't exist anymore.
link |
That wouldn't be true of a computer program.
link |
You could completely turn it off
link |
and a copy of it could be stored and you could recreate it.
link |
And so it has a different type of validity.
link |
You could actually take it back in time.
link |
You could eliminate its memory and have it go over again.
link |
I mean, it has a different kind of connotation
link |
Well, perhaps it can do the same thing with humans.
link |
It's just that we don't know how to do that yet.
link |
It's possible that we figure out all of these things
link |
on the machine first.
link |
But that doesn't mean the machine isn't conscious.
link |
I mean, if you look at the way people react,
link |
say, 3CPO or other machines that are conscious in movies,
link |
they don't actually present how it's conscious,
link |
but we see that they are a machine
link |
and people will believe that they are conscious
link |
and they'll actually worry about it
link |
if they get into trouble and so on.
link |
So 2029 is going to be the first year
link |
when a major thing happens.
link |
And that will shake our civilization
link |
to start to consider the role of AI in this world.
link |
I mean, this one guy at Google claimed
link |
that the machine was conscious.
link |
But that's just one person.
link |
When it starts to happen to scale.
link |
Well, that's exactly right because most people
link |
have not taken that position.
link |
I don't take that position.
link |
I mean, I've used different things like this
link |
and they don't appear to me to be conscious.
link |
As we eliminate various problems
link |
of these large language models,
link |
more and more people will accept that they're conscious.
link |
So when we get to 2029, I think a large fraction
link |
of people will believe that they're conscious.
link |
So it's not gonna happen all at once.
link |
I believe it will actually happen gradually
link |
and it's already started to happen.
link |
And so that takes us one step closer to the singularity.
link |
Another step then is in the 2030s
link |
when we can actually connect our neocortex,
link |
which is where we do our thinking, to computers.
link |
And I mean, just as this actually gains a lot
link |
to being connected to computers
link |
that will amplify its abilities,
link |
I mean, if this did not have any connection,
link |
it would be pretty stupid.
link |
It could not answer any of your questions.
link |
If you're just listening to this, by the way,
link |
Ray's holding up the all powerful smartphone.
link |
So we're gonna do that directly from our brains.
link |
I mean, these are pretty good.
link |
These already have amplified our intelligence.
link |
I'm already much smarter than I would otherwise be
link |
if I didn't have this.
link |
Because I remember my first book,
link |
The Age of Intelligent Machines,
link |
there was no way to get information from computers.
link |
I actually would go to a library, find a book,
link |
find the page that had an information I wanted,
link |
and I'd go to the copier,
link |
and my most significant information tool
link |
was a roll of quarters where I could feed the copier.
link |
So we're already greatly advanced
link |
that we have these things.
link |
There's a few problems with it.
link |
First of all, I constantly put it down,
link |
and I don't remember where I put it.
link |
I've actually never lost it.
link |
But you have to find it, and then you have to turn it on.
link |
So there's a certain amount of steps.
link |
It would actually be quite useful
link |
if someone would just listen to your conversation
link |
and say, oh, that's so and so actress,
link |
and tell you what you're talking about.
link |
So going from active to passive,
link |
where it just permeates your whole life.
link |
The way your brain does when you're awake.
link |
Your brain is always there.
link |
That's something that could actually
link |
just about be done today,
link |
where we'd listen to your conversation,
link |
understand what you're saying,
link |
understand what you're not missing,
link |
and give you that information.
link |
But another step is to actually go inside your brain.
link |
And there are some prototypes
link |
where you can connect your brain.
link |
They actually don't have the amount
link |
of bandwidth that we need.
link |
They can work, but they work fairly slowly.
link |
So if it actually would connect to your neocortex,
link |
and the neocortex, which I describe
link |
in How to Create a Mind,
link |
the neocortex is actually,
link |
it has different levels,
link |
and as you go up the levels,
link |
it's kind of like a pyramid.
link |
The top level is fairly small,
link |
and that's the level where you wanna connect
link |
these brain extenders.
link |
And so I believe that will happen in the 2030s.
link |
So just the way this is greatly amplified
link |
by being connected to the cloud,
link |
we can connect our own brain to the cloud,
link |
and just do what we can do by using this machine.
link |
Do you think it would look like
link |
the brain computer interface of like Neuralink?
link |
Well, Neuralink, it's an attempt to do that.
link |
It doesn't have the bandwidth that we need.
link |
Right, but I think,
link |
I mean, they're gonna get permission for this
link |
because there are a lot of people
link |
who absolutely need it because they can't communicate.
link |
I know a couple people like that
link |
who have ideas and they cannot,
link |
they cannot move their muscles and so on.
link |
They can't communicate.
link |
And so for them, this would be very valuable,
link |
but we could all use it.
link |
Basically, it'd be,
link |
turn us into something that would be like we have a phone,
link |
but it would be in our minds.
link |
It would be kind of instantaneous.
link |
And maybe communication between two people
link |
would not require this low bandwidth mechanism of language.
link |
We don't know what that would be,
link |
although we do know that computers can share information
link |
like language instantly.
link |
They can share many, many books in a second.
link |
So we could do that as well.
link |
If you look at what our brain does,
link |
it actually can manipulate different parameters.
link |
So we talk about these large language models.
link |
I mean, I had written that
link |
it requires a certain amount of information
link |
in order to be effective
link |
and that we would not see AI really being effective
link |
until it got to that level.
link |
And we had large language models
link |
that were like 10 billion bytes, didn't work very well.
link |
They finally got to a hundred billion bytes
link |
and now they work fairly well.
link |
And now we're going to a trillion bytes.
link |
If you say lambda has a hundred billion bytes,
link |
what does that mean?
link |
Well, what if you had something that had one byte,
link |
one parameter, maybe you wanna tell
link |
whether or not something's an elephant or not.
link |
And so you put in something that would detect its trunk.
link |
If it has a trunk, it's an elephant.
link |
If it doesn't have a trunk, it's not an elephant.
link |
That would work fairly well.
link |
There's a few problems with it.
link |
And it really wouldn't be able to tell what a trunk is,
link |
And maybe other things other than elephants have trunks,
link |
you might get really confused.
link |
I'm not sure which animals have trunks,
link |
but how do you define a trunk?
link |
But yeah, that's one parameter.
link |
So these things have a hundred billion parameters.
link |
So they're able to deal with very complex issues.
link |
All kinds of trunks.
link |
Human beings actually have a little bit more than that,
link |
but they're getting to the point
link |
where they can emulate humans.
link |
If we were able to connect this to our neocortex,
link |
we would basically add more of these abilities
link |
to make distinctions,
link |
and it could ultimately be much smarter
link |
and also be attached to information
link |
that we feel is reliable.
link |
So that's where we're headed.
link |
So you think that there will be a merger in the 30s,
link |
an increasing amount of merging
link |
between the human brain and the AI brain?
link |
And the AI brain is really an emulation of human beings.
link |
I mean, that's why we're creating them,
link |
because human beings act the same way,
link |
and this is basically to amplify them.
link |
I mean, this amplifies our brain.
link |
It's a little bit clumsy to interact with,
link |
but it definitely is way beyond what we had 15 years ago.
link |
But the implementation becomes different,
link |
just like a bird versus the airplane,
link |
even though the AI brain is an emulation,
link |
it starts adding features we might not otherwise have,
link |
like ability to consume a huge amount
link |
of information quickly,
link |
like look up thousands of Wikipedia articles in one take.
link |
I mean, we can get, for example,
link |
issues like simulated biology,
link |
where it can simulate many different things at once.
link |
We already had one example of simulated biology,
link |
which is the Moderna vaccine.
link |
And that's gonna be now
link |
the way in which we create medications.
link |
But they were able to simulate
link |
what each example of an mRNA would do to a human being,
link |
and they were able to simulate that quite reliably.
link |
And we actually simulated billions
link |
of different mRNA sequences,
link |
and they found the ones that were the best,
link |
and they created the vaccine.
link |
And they did, and talked about doing that quickly,
link |
they did that in two days.
link |
Now, how long would a human being take
link |
to simulate billions of different mRNA sequences?
link |
I don't know that we could do it at all,
link |
but it would take many years.
link |
They did it in two days, and one of the reasons
link |
that people didn't like vaccines
link |
is because it was done too quickly,
link |
it was done too fast.
link |
And they actually included the time it took to test it out,
link |
which was 10 months, so they figured,
link |
okay, it took 10 months to create this.
link |
Actually, it took us two days.
link |
And we also will be able to ultimately do the tests
link |
in a few days as well.
link |
Oh, because we can simulate how the body will respond to it.
link |
Yeah, that's a little bit more complicated
link |
because the body has a lot of different elements,
link |
and we have to simulate all of that,
link |
but that's coming as well.
link |
So ultimately, we could create it in a few days
link |
and then test it in a few days, and it would be done.
link |
And we can do that with every type
link |
of medical insufficiency that we have.
link |
So curing all diseases, improving certain functions
link |
of the body, supplements, drugs for recreation,
link |
for health, for performance, for productivity,
link |
all that kind of stuff.
link |
Well, that's where we're headed,
link |
because I mean, right now we have a very inefficient way
link |
of creating these new medications.
link |
But we've already shown it, and the Moderna vaccine
link |
is actually the best of the vaccines we've had,
link |
and it literally took two days to create.
link |
And we'll get to the point
link |
where we can test it out also quickly.
link |
Are you impressed by AlphaFold
link |
and the solution to the protein folding,
link |
which essentially is simulating, modeling
link |
this primitive building block of life,
link |
which is a protein, and its 3D shape?
link |
It's pretty remarkable that they can actually predict
link |
what the 3D shape of these things are,
link |
but they did it with the same type of neural net
link |
that won, for example, the Go test.
link |
So it's all the same.
link |
It's all the same.
link |
All the same approaches.
link |
They took that same thing and just changed the rules
link |
to chess, and within a couple of days,
link |
it now played a master level of chess
link |
greater than any human being.
link |
And the same thing then worked for AlphaFold,
link |
which no human had done.
link |
I mean, human beings could do,
link |
the best humans could maybe do 15, 20%
link |
of figuring out what the shape would be.
link |
And after a few takes, it ultimately did just about 100%.
link |
Do you still think the singularity will happen in 2045?
link |
And what does that look like?
link |
Once we can amplify our brain with computers directly,
link |
which will happen in the 2030s,
link |
that's gonna keep growing.
link |
That's another whole theme,
link |
which is the exponential growth of computing power.
link |
Yeah, so looking at price performance of computation
link |
from 1939 to 2021.
link |
Right, so that starts with the very first computer
link |
actually created by a German during World War II.
link |
You might have thought that that might be significant,
link |
but actually the Germans didn't think computers
link |
were significant, and they completely rejected it.
link |
The second one is also the ZUSA 2.
link |
And by the way, we're looking at a plot
link |
with the X axis being the year from 1935 to 2025.
link |
And on the Y axis in log scale
link |
is computation per second per constant dollar.
link |
So dollar normalized inflation.
link |
And it's growing linearly on the log scale,
link |
which means it's growing exponentially.
link |
The third one was the British computer,
link |
which the Allies did take very seriously.
link |
And it cracked the German code
link |
and enables the British to win the Battle of Britain,
link |
which otherwise absolutely would not have happened
link |
if they hadn't cracked the code using that computer.
link |
But that's an exponential graph.
link |
So a straight line on that graph is exponential growth.
link |
And you see 80 years of exponential growth.
link |
And I would say about every five years,
link |
and this happened shortly before the pandemic,
link |
people saying, well, they call it Moore's law,
link |
which is not the correct, because that's not all intel.
link |
In fact, this started decades before intel was even created.
link |
It wasn't with transistors formed into a grid.
link |
So it's not just transistor count or transistor size.
link |
Right, it started with relays, then went to vacuum tubes,
link |
then went to individual transistors,
link |
and then to integrated circuits.
link |
And integrated circuits actually starts
link |
like in the middle of this graph.
link |
And it has nothing to do with intel.
link |
Intel actually was a key part of this.
link |
But a few years ago, they stopped making the fastest chips.
link |
But if you take the fastest chip of any technology
link |
in that year, you get this kind of graph.
link |
And it's definitely continuing for 80 years.
link |
So you don't think Moore's law, broadly defined, is dead.
link |
It's been declared dead multiple times throughout this process.
link |
I don't like the term Moore's law,
link |
because it has nothing to do with Moore or with intel.
link |
But yes, the exponential growth of computing is continuing.
link |
It has never stopped.
link |
From various sources.
link |
I mean, it went through World War II,
link |
it went through global recessions.
link |
It's just continuing.
link |
And if you continue that out, along with software gains,
link |
which is a whole nother issue,
link |
and they really multiply,
link |
whatever you get from software gains,
link |
you multiply by the computer gains,
link |
you get faster and faster speed.
link |
This is actually the fastest computer models
link |
that have been created.
link |
And that actually expands roughly twice a year.
link |
Like, every six months it expands by two.
link |
So we're looking at a plot from 2010 to 2022.
link |
On the x axis is the publication date of the model,
link |
and perhaps sometimes the actual paper associated with it.
link |
And on the y axis is training, compute, and flops.
link |
And so basically this is looking at the increase
link |
in the, not transistors,
link |
but the computational power of neural networks.
link |
Yes, the computational power that created these models.
link |
And that's doubled every six months.
link |
Which is even faster than transistor division.
link |
Now actually, since it goes faster than the amount of cost,
link |
this has actually become a greater investment
link |
But at any rate, by the time we get to 2045,
link |
we'll be able to multiply our intelligence
link |
many millions fold.
link |
And it's just very hard to imagine what that will be like.
link |
And that's the singularity where we can't even imagine.
link |
Right, that's why we call it the singularity.
link |
Because the singularity in physics,
link |
something gets sucked into its singularity
link |
and you can't tell what's going on in there
link |
because no information can get out of it.
link |
There's various problems with that,
link |
but that's the idea.
link |
It's too much beyond what we can imagine.
link |
Do you think it's possible we don't notice
link |
that what the singularity actually feels like
link |
is we just live through it
link |
with exponentially increasing cognitive capabilities
link |
and we almost, because everything's moving so quickly,
link |
aren't really able to introspect
link |
that our life has changed.
link |
Yeah, but I mean, we will have that much greater capacity
link |
to understand things, so we should be able to look back.
link |
Looking at history, understand history.
link |
But we will need people, basically like you and me,
link |
to actually think about these things.
link |
But we might be distracted
link |
by all the other sources of entertainment and fun
link |
because the exponential power of intellect is growing,
link |
but also there'll be a lot of fun.
link |
The amount of ways you can have, you know.
link |
I mean, we already have a lot of fun with computer games
link |
and so on that are really quite remarkable.
link |
What do you think about the digital world,
link |
the metaverse, virtual reality?
link |
Will that have a component in this
link |
or will most of our advancement be in physical reality?
link |
Well, that's a little bit like Second Life,
link |
although the Second Life actually didn't work very well
link |
because it couldn't actually handle too many people.
link |
And I don't think the metaverse has come to being.
link |
I think there will be something like that.
link |
It won't necessarily be from that one company.
link |
I mean, there's gonna be competitors.
link |
But yes, we're gonna live increasingly online,
link |
and particularly if our brains are online.
link |
I mean, how could we not be online?
link |
Do you think it's possible that given this merger with AI,
link |
most of our meaningful interactions
link |
will be in this virtual world most of our life?
link |
We fall in love, we make friends,
link |
we come up with ideas, we do collaborations, we have fun.
link |
I actually know somebody who's marrying somebody
link |
that they never met.
link |
I think they just met her briefly before the wedding,
link |
but she actually fell in love with this other person,
link |
never having met them.
link |
And I think the love is real, so.
link |
That's a beautiful story,
link |
but do you think that story is one that might be experienced
link |
as opposed to by hundreds of thousands of people,
link |
but instead by hundreds of millions of people?
link |
I mean, it really gives you appreciation
link |
for these virtual ways of communicating.
link |
And if anybody can do it,
link |
then it's really not such a freak story.
link |
So I think more and more people will do that.
link |
But that's turning our back
link |
on our entire history of evolution.
link |
The old days, we used to fall in love by holding hands
link |
and sitting by the fire, that kind of stuff.
link |
Here, you're playing.
link |
Actually, I have five patents on where you can hold hands,
link |
even if you're separated.
link |
So the touch, the sense, it's all just senses.
link |
It's all just replicated.
link |
Yeah, I mean, touch is,
link |
it's not just that you're touching someone or not.
link |
There's a whole way of doing it, and it's very subtle.
link |
But ultimately, we can emulate all of that.
link |
Are you excited by that future?
link |
Do you worry about that future?
link |
I have certain worries about the future,
link |
but not virtual touch.
link |
Well, I agree with you.
link |
You described six stages
link |
in the evolution of information processing in the universe,
link |
as you started to describe.
link |
Can you maybe talk through some of those stages
link |
from the physics and chemistry to DNA and brains,
link |
and then to the very end,
link |
to the very beautiful end of this process?
link |
It actually gets more rapid.
link |
So physics and chemistry, that's how we started.
link |
So the very beginning of the universe.
link |
We had lots of electrons and various things traveling around.
link |
And that took actually many billions of years,
link |
kind of jumping ahead here to kind of
link |
some of the last stages where we have things
link |
like love and creativity.
link |
It's really quite remarkable that that happens.
link |
But finally, physics and chemistry created biology and DNA.
link |
And now you had actually one type of molecule
link |
that described the cutting edge of this process.
link |
And we go from physics and chemistry to biology.
link |
And finally, biology created brains.
link |
I mean, not everything that's created by biology
link |
has a brain, but eventually brains came along.
link |
And all of this is happening faster and faster.
link |
It created increasingly complex organisms.
link |
Another key thing is actually not just brains,
link |
Because there's a lot of animals
link |
with brains even bigger than humans.
link |
I mean, elephants have a bigger brain.
link |
Whales have a bigger brain.
link |
But they've not created technology
link |
because they don't have a thumb.
link |
So that's one of the really key elements
link |
in the evolution of humans.
link |
This physical manipulator device
link |
that's useful for puzzle solving in the physical reality.
link |
So I could think, I could look at a tree and go,
link |
oh, I could actually trip that branch down
link |
and eliminate the leaves and carve a tip on it
link |
and I would create technology.
link |
And you can't do that if you don't have a thumb.
link |
So thumbs then created technology
link |
and technology also had a memory.
link |
And now those memories are competing
link |
with the scale and scope of human beings.
link |
And ultimately we'll go beyond it.
link |
And then we're gonna merge human technology
link |
with human intelligence
link |
and understand how human intelligence works,
link |
which I think we already do.
link |
And we're putting that into our human technology.
link |
So create the technology inspired by our own intelligence
link |
and then that technology supersedes us
link |
in terms of its capabilities.
link |
And we ride along.
link |
Or do you ultimately see it as...
link |
And we ride along, but a lot of people don't see that.
link |
They say, well, you've got humans and you've got machines
link |
and there's no way we can ultimately compete with humans.
link |
And you can already see that.
link |
Lee Soudal, who's like the best Go player in the world,
link |
says he's not gonna play Go anymore.
link |
Because playing Go for a human,
link |
that was like the ultimate in intelligence
link |
because no one else could do that.
link |
But now a machine can actually go way beyond him.
link |
And so he says, well, there's no point playing it anymore.
link |
That may be more true for games than it is for life.
link |
I think there's a lot of benefit
link |
to working together with AI in regular life.
link |
So if you were to put a probability on it,
link |
is it more likely that we merge with AI
link |
or AI replaces us?
link |
A lot of people just think computers come along
link |
and they compete with them.
link |
We can't really compete and that's the end of it.
link |
As opposed to them increasing our abilities.
link |
And if you look at most technology,
link |
it increases our abilities.
link |
I mean, look at the history of work.
link |
Look at what people did 100 years ago.
link |
Does any of that exist anymore?
link |
People, I mean, if you were to predict
link |
that all of these jobs would go away
link |
and would be done by machines,
link |
people would say, well, there's gonna be,
link |
no one's gonna have jobs
link |
and it's gonna be massive unemployment.
link |
But I show in this book that's coming out
link |
the amount of people that are working,
link |
even as a percentage of the population has gone way up.
link |
We're looking at the x axis year from 1774 to 2024
link |
and on the y axis, personal income per capita
link |
in constant dollars and it's growing super linearly.
link |
I mean, it's 2021 constant dollars and it's gone way up.
link |
That's not what you would predict
link |
given that we would predict
link |
that all these jobs would go away.
link |
But the reason it's gone up is because
link |
we've basically enhanced our own capabilities
link |
by using these machines
link |
as opposed to them just competing with us.
link |
That's a key way in which we're gonna be able
link |
to become far smarter than we are now
link |
by increasing the number of different parameters
link |
we can consider in making a decision.
link |
I was very fortunate, I am very fortunate
link |
to be able to get a glimpse preview
link |
of your upcoming book, Singularity is Nearer.
link |
And one of the themes outside of just discussing
link |
the increasing exponential growth of technology,
link |
one of the themes is that things are getting better
link |
in all aspects of life.
link |
And you talked just about this.
link |
So one of the things you're saying is with jobs.
link |
So let me just ask about that.
link |
There is a big concern that automation,
link |
especially powerful AI, will get rid of jobs.
link |
There are people who lose jobs.
link |
And as you were saying, the sense is
link |
throughout the history of the 20th century,
link |
automation did not do that ultimately.
link |
And so the question is, will this time be different?
link |
Right, that is the question.
link |
Will this time be different?
link |
And it really has to do with how quickly
link |
we can merge with this type of intelligence.
link |
Whether Lambda or GPT3 is out there,
link |
and maybe it's overcome some of its key problems,
link |
and we really haven't enhanced human intelligence,
link |
that might be a negative scenario.
link |
But I mean, that's why we create technologies,
link |
to enhance ourselves.
link |
And I believe we will be enhanced
link |
when I'm just going to sit here with
link |
300 million modules in our neocortex.
link |
We're going to be able to go beyond that.
link |
Because that's useful, but we can multiply that by 10,
link |
100, 1,000, a million.
link |
And you might think, well, what's the point of doing that?
link |
It's like asking somebody that's never heard music,
link |
well, what's the value of music?
link |
I mean, you can't appreciate it until you've created it.
link |
There's some worry that there'll be a wealth disparity.
link |
Class or wealth disparity, only the rich people
link |
will be, basically, the rich people
link |
will first have access to this kind of thing,
link |
and then because of this kind of thing,
link |
because the ability to merge
link |
will get richer exponentially faster.
link |
And I say that's just like cell phones.
link |
I mean, there's like four billion cell phones
link |
in the world today.
link |
In fact, when cell phones first came out,
link |
you had to be fairly wealthy.
link |
They weren't very inexpensive.
link |
So you had to have some wealth in order to afford them.
link |
Yeah, there were these big, sexy phones.
link |
And they didn't work very well.
link |
They did almost nothing.
link |
So you can only afford these things if you're wealthy
link |
at a point where they really don't work very well.
link |
So achieving scale and making it inexpensive
link |
is part of making the thing work well.
link |
So these are not totally cheap, but they're pretty cheap.
link |
I mean, you can get them for a few hundred dollars.
link |
Especially given the kind of things it provides for you.
link |
There's a lot of people in the third world
link |
that have very little, but they have a smartphone.
link |
And the same will be true with AI.
link |
I mean, I see homeless people have their own cell phones.
link |
Yeah, so your sense is any kind of advanced technology
link |
will take the same trajectory.
link |
Right, it ultimately becomes cheap and will be affordable.
link |
I probably would not be the first person
link |
to put something in my brain to connect to computers
link |
because I think it will have limitations.
link |
But once it's really perfected,
link |
and at that point it'll be pretty inexpensive,
link |
I think it'll be pretty affordable.
link |
So in which other ways, as you outline your book,
link |
is life getting better?
link |
Because I think...
link |
Well, I mean, I have 50 charts in there
link |
where everything is getting better.
link |
I think there's a kind of cynicism about,
link |
like even if you look at extreme poverty, for example.
link |
For example, this is actually a poll
link |
taken on extreme poverty, and people were asked,
link |
has poverty gotten better or worse?
link |
And the options are increased by 50%,
link |
increased by 25%, remain the same,
link |
decreased by 25%, decreased by 50%.
link |
If you're watching this or listening to this,
link |
try to vote for yourself.
link |
70% thought it had gotten worse,
link |
and that's the general impression.
link |
88% thought it had gotten worse or remained the same.
link |
Only 1% thought it decreased by 50%,
link |
and that is the answer.
link |
It actually decreased by 50%.
link |
So only 1% of people got the right optimistic estimate
link |
of how poverty is.
link |
Right, and this is the reality,
link |
and it's true of almost everything you look at.
link |
You don't wanna go back 100 years or 50 years.
link |
Things were quite miserable then,
link |
but we tend not to remember that.
link |
So literacy rate increasing over the past few centuries
link |
across all the different nations,
link |
nearly to 100% across many of the nations in the world.
link |
Average years of education have gone way up.
link |
Life expectancy is also increasing.
link |
Life expectancy was 48 in 1900.
link |
And it's over 80 now.
link |
And it's gonna continue to go up,
link |
particularly as we get into more advanced stages
link |
of simulated biology.
link |
For life expectancy, these trends are the same
link |
for at birth, age one, age five, age 10,
link |
so it's not just the infant mortality.
link |
And I have 50 more graphs in the book
link |
about all kinds of things.
link |
Even spread of democracy,
link |
which might bring up some sort of controversial issues,
link |
it still has gone way up.
link |
Well, that one has gone way up,
link |
but that one is a bumpy road, right?
link |
Exactly, and somebody might represent democracy
link |
and go backwards, but we basically had no democracies
link |
before the creation of the United States,
link |
which was a little over two centuries ago,
link |
which in the scale of human history isn't that long.
link |
Do you think superintelligence systems will help
link |
So what is democracy?
link |
Democracy is giving a voice to the populace
link |
and having their ideas, having their beliefs,
link |
having their views represented.
link |
I mean, we've seen social networks
link |
can spread conspiracy theories,
link |
which have been quite negative,
link |
being, for example, being against any kind of stuff
link |
that would help your health.
link |
So those kinds of ideas have,
link |
on social media, what you notice is they increase
link |
engagement, so dramatic division increases engagement.
link |
Do you worry about AI systems that will learn
link |
to maximize that division?
link |
I mean, I do have some concerns about this,
link |
and I have a chapter in the book about the perils
link |
of advanced AI, spreading misinformation
link |
on social networks is one of them,
link |
but there are many others.
link |
What's the one that worries you the most
link |
that we should think about to try to avoid?
link |
Well, it's hard to choose.
link |
We do have the nuclear power that evolved
link |
when I was a child, I remember,
link |
and we would actually do these drills against a nuclear war.
link |
We'd get under our desks and put our hands behind our heads
link |
to protect us from a nuclear war.
link |
Seems to work, we're still around, so.
link |
But that's still a concern.
link |
And there are key dangerous situations
link |
that can take place in biology.
link |
Someone could create a virus that's very,
link |
I mean, we have viruses that are hard to spread,
link |
and they can be very dangerous,
link |
and we have viruses that are easy to spread,
link |
but they're not so dangerous.
link |
Somebody could create something
link |
that would be very easy to spread and very dangerous,
link |
and be very hard to stop.
link |
It could be something that would spread
link |
without people noticing, because people could get it,
link |
they'd have no symptoms, and then everybody would get it,
link |
and then symptoms would occur maybe a month later.
link |
So I mean, and that actually doesn't occur normally,
link |
because if we were to have a problem with that,
link |
we wouldn't exist.
link |
So the fact that humans exist means that we don't have
link |
viruses that can spread easily and kill us,
link |
because otherwise we wouldn't exist.
link |
Yeah, viruses don't wanna do that.
link |
They want to spread and keep the host alive somewhat.
link |
So you can describe various dangers with biology.
link |
Also nanotechnology, which we actually haven't experienced
link |
yet, but there are people that are creating nanotechnology,
link |
and I describe that in the book.
link |
Now you're excited by the possibilities of nanotechnology,
link |
of nanobots, of being able to do things inside our body,
link |
inside our mind, that's going to help.
link |
What's exciting, what's terrifying about nanobots?
link |
What's exciting is that that's a way to communicate
link |
with our neocortex, because each neocortex is pretty small
link |
and you need a small entity that can actually get in there
link |
and establish a communication channel.
link |
And that's gonna really be necessary to connect our brains
link |
to AI within ourselves, because otherwise it would be hard
link |
for us to compete with it.
link |
In a high bandwidth way.
link |
And that's key, actually, because a lot of the things
link |
like Neuralink are really not high bandwidth yet.
link |
So nanobots is the way you achieve high bandwidth.
link |
How much intelligence would those nanobots have?
link |
Yeah, they don't need a lot, just enough to basically
link |
establish a communication channel to one nanobot.
link |
So it's primarily about communication.
link |
Between external computing devices
link |
and our biological thinking machine.
link |
What worries you about nanobots?
link |
Is it similar to with the viruses?
link |
Well, I mean, it's the great goo challenge.
link |
If you had a nanobot that wanted to create
link |
any kind of entity and repeat itself,
link |
and was able to operate in a natural environment,
link |
it could turn everything into that entity
link |
and basically destroy all biological life.
link |
So you mentioned nuclear weapons.
link |
I'd love to hear your opinion about the 21st century
link |
and whether you think we might destroy ourselves.
link |
And maybe your opinion, if it has changed
link |
by looking at what's going on in Ukraine,
link |
that we could have a hot war with nuclear powers involved
link |
and the tensions building and the seeming forgetting
link |
of how terrifying and destructive nuclear weapons are.
link |
Do you think humans might destroy ourselves
link |
in the 21st century, and if we do, how?
link |
And how do we avoid it?
link |
I don't think that's gonna happen
link |
despite the terrors of that war.
link |
It is a possibility, but I mean, I don't.
link |
It's unlikely in your mind.
link |
Yeah, even with the tensions we've had
link |
with this one nuclear power plant that's been taken over,
link |
it's very tense, but I don't actually see a lot of people
link |
worrying that that's gonna happen.
link |
I think we'll avoid that.
link |
We had two nuclear bombs go off in 45,
link |
so now we're 77 years later.
link |
Yeah, we're doing pretty good.
link |
We've never had another one go off through anger.
link |
People forget the lessons of history.
link |
Well, yeah, I mean, I am worried about it.
link |
I mean, that is definitely a challenge.
link |
But you believe that we'll make it out
link |
and ultimately superintelligent AI will help us make it out
link |
as opposed to destroy us.
link |
I think so, but we do have to be mindful of these dangers.
link |
And there are other dangers besides nuclear weapons, so.
link |
So to get back to merging with AI,
link |
will we be able to upload our mind in a computer
link |
in a way where we might even transcend
link |
the constraints of our bodies?
link |
So copy our mind into a computer and leave the body behind?
link |
Let me describe one thing I've already done with my father.
link |
That's a great story.
link |
So we created a technology, this is public,
link |
came out, I think, six years ago,
link |
where you could ask any question
link |
and the release product,
link |
which I think is still on the market,
link |
it would read 200,000 books.
link |
And then find the one sentence in 200,000 books
link |
that best answered your question.
link |
And it's actually quite interesting.
link |
You can ask all kinds of questions
link |
and you get the best answer in 200,000 books.
link |
But I was also able to take it
link |
and not go through 200,000 books,
link |
but go through a book that I put together,
link |
which is basically everything my father had written.
link |
So everything he had written, I had gathered,
link |
and we created a book,
link |
everything that Frederick Herzog had written.
link |
Now, I didn't think this actually would work that well
link |
because stuff he had written was stuff about how to lay out.
link |
I mean, he directed choral groups
link |
and he would be laying out how the people should,
link |
where they should sit and how to fund this
link |
and all kinds of things
link |
that really didn't seem that interesting.
link |
And yet, when you ask a question,
link |
it would go through it
link |
and it would actually give you a very good answer.
link |
So I said, well, who's the most interesting composer?
link |
And he said, well, definitely Brahms.
link |
And he would go on about how Brahms was fabulous
link |
and talk about the importance of music education.
link |
So you could have essentially a question and answer,
link |
a conversation with him.
link |
You could have a conversation with him,
link |
which was actually more interesting than talking to him
link |
because if you talked to him,
link |
he'd be concerned about how they're gonna lay out
link |
this property to give a choral group.
link |
He'd be concerned about the day to day
link |
versus the big questions.
link |
And you did ask about the meaning of life
link |
and he answered, love.
link |
Yeah, you get used to missing somebody after 52 years,
link |
and I didn't really have intelligent conversations with him
link |
until later in life.
link |
In the last few years, he was sick,
link |
which meant he was home a lot
link |
and I was actually able to talk to him
link |
about different things like music and other things.
link |
And so I miss that very much.
link |
What did you learn about life from your father?
link |
What part of him is with you now?
link |
He was devoted to music.
link |
And when he would create something to music,
link |
it put him in a different world.
link |
Otherwise, he was very shy.
link |
And if people got together,
link |
he tended not to interact with people
link |
just because of his shyness.
link |
But when he created music, he was like a different person.
link |
Do you have that in you?
link |
That kind of light that shines?
link |
I mean, I got involved with technology at like age five.
link |
And you fell in love with it
link |
in the same way he did with music?
link |
I remember this actually happened with my grandmother.
link |
She had a manual typewriter
link |
and she wrote a book, One Life Is Not Enough,
link |
which actually a good title for a book I might write,
link |
but it was about a school she had created.
link |
Well, actually her mother created it.
link |
So my mother's mother's mother created the school in 1868.
link |
And it was the first school in Europe
link |
that provided higher education for girls.
link |
It went through 14th grade.
link |
If you were a girl and you were lucky enough
link |
to get an education at all,
link |
it would go through like ninth grade.
link |
And many people didn't have any education as a girl.
link |
This went through 14th grade.
link |
Her mother created it, she took it over,
link |
and the book was about the history of the school
link |
and her involvement with it.
link |
When she presented it to me,
link |
I was not so interested in the story of the school,
link |
but I was totally amazed with this manual typewriter.
link |
I mean, here is something you could put a blank piece
link |
of paper into and you could turn it into something
link |
that looked like it came from a book.
link |
And you can actually type on it
link |
and it looked like it came from a book.
link |
It was just amazing to me.
link |
And I could see actually how it worked.
link |
And I was also interested in magic.
link |
But in magic, if somebody actually knows how it works,
link |
the magic goes away.
link |
The magic doesn't stay there
link |
if you actually understand how it works.
link |
But here was technology.
link |
I didn't have that word when I was five or six.
link |
And the magic was still there for you?
link |
The magic was still there, even if you knew how it worked.
link |
So I became totally interested in this
link |
and then went around, collected little pieces
link |
of mechanical objects from bicycles, from broken radios.
link |
I would go through the neighborhood.
link |
This was an era where you would allow five or six year olds
link |
to run through the neighborhood and do this.
link |
We don't do that anymore.
link |
But I didn't know how to put them together.
link |
I said, if I could just figure out
link |
how to put these things together, I could solve any problem.
link |
And I actually remember talking to these very old girls.
link |
I think they were 10.
link |
And telling them, if I could just figure this out,
link |
we could fly, we could do anything.
link |
And they said, well, you have quite an imagination.
link |
And then when I was in third grade,
link |
so I was like eight,
link |
created like a virtual reality theater
link |
where people could come on stage
link |
and they could move their arms.
link |
And all of it was controlled through one control box.
link |
It was all done with mechanical technology.
link |
And it was a big hit in my third grade class.
link |
And then I went on to do things
link |
in junior high school science fairs
link |
and high school science fairs.
link |
I won the Westinghouse Science Talent Search.
link |
So I mean, I became committed to technology
link |
when I was five or six years old.
link |
You've talked about how you use lucid dreaming to think,
link |
to come up with ideas as a source of creativity.
link |
Because you maybe talk through that,
link |
maybe the process of how to,
link |
you've invented a lot of things.
link |
You've came up and thought through
link |
some very interesting ideas.
link |
What advice would you give,
link |
or can you speak to the process of thinking,
link |
of how to think, how to think creatively?
link |
Well, I mean, sometimes I will think through in a dream
link |
and try to interpret that.
link |
But I think the key issue that I would tell younger people
link |
is to put yourself in the position
link |
that what you're trying to create already exists.
link |
And then you're explaining, like...
link |
That's really interesting.
link |
You paint a world that you would like to exist,
link |
you think it exists, and reverse engineer that.
link |
And then you actually imagine you're giving a speech
link |
about how you created this.
link |
Well, you'd have to then work backwards
link |
as to how you would create it in order to make it work.
link |
And that requires some imagination too,
link |
some first principles thinking.
link |
You have to visualize that world.
link |
That's really interesting.
link |
And generally, when I talk about things
link |
we're trying to invent, I would use the present tense
link |
as if it already exists.
link |
Not just to give myself that confidence,
link |
but everybody else who's working on it.
link |
We just have to kind of do all the steps
link |
in order to make it actual.
link |
How much of a good idea is about timing?
link |
How much is it about your genius
link |
versus that its time has come?
link |
Timing's very important.
link |
I mean, that's really why I got into futurism.
link |
I didn't, I wasn't inherently a futurist.
link |
That was not really my goal.
link |
It's really to figure out when things are feasible.
link |
We see that now with large scale models.
link |
The very large scale models like GPT3,
link |
it started two years ago.
link |
Four years ago, it wasn't feasible.
link |
In fact, they did create GPT2, which didn't work.
link |
So it required a certain amount of timing
link |
having to do with this exponential growth
link |
of computing power.
link |
So futurism in some sense is a study of timing,
link |
trying to understand how the world will evolve
link |
and when will the capacity for certain ideas emerge.
link |
And that's become a thing in itself
link |
and to try to time things in the future.
link |
But really its original purpose was to time my products.
link |
I mean, I did OCR in the 1970s
link |
because OCR doesn't require a lot of computation.
link |
Optical character recognition.
link |
Yeah, so we were able to do that in the 70s
link |
and I waited till the 80s to address speech recognition
link |
since that requires more computation.
link |
So you were thinking through timing
link |
when you're developing those things.
link |
And that's how you've developed that brain power
link |
to start to think in a futurist sense
link |
when how will the world look like in 2045
link |
and work backwards and how it gets there.
link |
But that has to become a thing in itself
link |
because looking at what things will be like in the future
link |
and the future reflects such dramatic changes in how humans will live
link |
that was worth communicating also.
link |
So you developed that muscle of predicting the future
link |
and then applied broadly
link |
and started to discuss how it changes the world of technology,
link |
how it changes the world of human life on earth.
link |
In Danielle, one of your books,
link |
you write about someone who has the courage
link |
to question assumptions that limit human imagination
link |
to solve problems.
link |
And you also give advice
link |
on how each of us can have this kind of courage.
link |
Well, it's good that you picked that quote
link |
because I think that does symbolize what Danielle is about.
link |
So how can each of us have that courage
link |
to question assumptions?
link |
I mean, we see that when people can go beyond
link |
the current realm and create something that's new.
link |
I mean, take Uber, for example.
link |
Before that existed, you never thought
link |
that that would be feasible
link |
and it did require changes in the way people work.
link |
Is there practical advice as you give in the book
link |
about what each of us can do to be a Danielle?
link |
Well, she looks at the situation
link |
and tries to imagine how she can overcome various obstacles
link |
and then she goes for it.
link |
And she's a very good communicator
link |
so she can communicate these ideas to other people.
link |
And there's practical advice of learning to program
link |
and recording your life and things of this nature.
link |
Become a physicist.
link |
So you list a bunch of different suggestions
link |
of how to throw yourself into this world.
link |
Yeah, I mean, it's kind of an idea
link |
how young people can actually change the world
link |
by learning all of these different skills.
link |
And at the core of that is the belief
link |
that you can change the world.
link |
That your mind, your body can change the world.
link |
Yeah, that's right.
link |
And not letting anyone else tell you otherwise.
link |
That's really good, exactly.
link |
When we upload the story you told about your dad
link |
and having a conversation with him,
link |
we're talking about uploading your mind to the computer.
link |
Do you think we'll have a future
link |
with something you call afterlife?
link |
We'll have avatars that mimic increasingly better and better
link |
our behavior, our appearance, all that kind of stuff.
link |
Even those that are perhaps no longer with us.
link |
Yes, I mean, we need some information about them.
link |
I mean, think about my father.
link |
I have what he wrote.
link |
Now, he didn't have a word processor,
link |
so he didn't actually write that much.
link |
And our memories of him aren't perfect.
link |
So how do you even know if you've created something
link |
that's satisfactory?
link |
Now, you could do a Frederick Kurzweil Turing test.
link |
It seems like Frederick Kurzweil to me.
link |
But the people who remember him, like me,
link |
don't have a perfect memory.
link |
Is there such a thing as a perfect memory?
link |
Maybe the whole point is for him to make you feel
link |
Yeah, well, I think that would be the goal.
link |
And that's the connection we have with loved ones.
link |
It's not really based on very strict definition of truth.
link |
It's more about the experiences we share.
link |
And they get morphed through memory.
link |
But ultimately, they make us smile.
link |
I think we definitely can do that.
link |
And that would be very worthwhile.
link |
So do you think we'll have a world of replicants?
link |
There'll be a bunch of Ray Kurzweils.
link |
Like, I could hang out with one.
link |
I can download it for five bucks
link |
and have a best friend, Ray.
link |
And you, the original copy, wouldn't even know about it.
link |
Is that, do you think that world is,
link |
first of all, do you think that world is feasible?
link |
And do you think there's ethical challenges there?
link |
Like, how would you feel about me hanging out
link |
with Ray Kurzweil and you not knowing about it?
link |
It doesn't strike me as a problem.
link |
Which you, the original?
link |
Would you strike, would that cause a problem for you?
link |
No, I would really very much enjoy it.
link |
No, not just hang out with me,
link |
but if somebody hang out with you, a replicant of you.
link |
Well, I think I would start, it sounds exciting,
link |
but then what if they start doing better than me
link |
and take over my friend group?
link |
And then, because they may be an imperfect copy
link |
or there may be more social, all these kinds of things,
link |
and then I become like the old version
link |
that's not nearly as exciting.
link |
Maybe they're a copy of the best version of me
link |
Yeah, but if you hang out with a replicant of me
link |
and that turned out to be successful,
link |
I'd feel proud of that person because it was based on me.
link |
So it's, but it is a kind of death of this version of you.
link |
Well, not necessarily.
link |
I mean, you can still be alive, right?
link |
But, and you would be proud, okay,
link |
so it's like having kids and you're proud
link |
that they've done even more than you were able to do.
link |
It does bring up new issues,
link |
but it seems like an opportunity.
link |
Well, that replicant should probably have the same rights
link |
Well, that gets into a whole issue
link |
because when a replicant occurs,
link |
they're not necessarily gonna have your rights.
link |
And if a replicant occurs,
link |
if it's somebody who's already dead,
link |
do they have all the obligations
link |
and that the original person had?
link |
Do they have all the agreements that they had?
link |
I think you're gonna have to have laws that say yes.
link |
There has to be, if you wanna create a replicant,
link |
they have to have all the same rights as human rights.
link |
Well, you don't know.
link |
Someone can create a replicant and say,
link |
well, it's a replicant,
link |
but I didn't bother getting their rights.
link |
Yeah, but that would be illegal, I mean.
link |
Like if you do that, you have to do that in the black market.
link |
If you wanna get an official replicant.
link |
Okay, it's not so easy.
link |
It's supposed to create multiple replicants.
link |
The original rights,
link |
maybe for one person and not for a whole group of people.
link |
So there has to be at least one.
link |
And then all the other ones kinda share the rights.
link |
Yeah, I just don't think that,
link |
that's very difficult to conceive for us humans,
link |
the idea that this country.
link |
You create a replicant that has certain,
link |
I mean, I've talked to people about this,
link |
including my wife, who would like to get back her father.
link |
And she doesn't worry about who has rights to what.
link |
She would have somebody that she could visit with
link |
and might give her some satisfaction.
link |
And she wouldn't care about any of these other rights.
link |
What does your wife think about multiple rake or as wells?
link |
Have you had that discussion?
link |
I haven't addressed that with her.
link |
I think ultimately that's an important question,
link |
loved ones, how they feel about.
link |
There's something about love.
link |
Well, that's the key thing, right?
link |
If the loved one's rejected,
link |
it's not gonna work very well, so.
link |
So the loved ones really are the key determinant,
link |
whether or not this works or not.
link |
But there's also ethical rules.
link |
We have to contend with the idea,
link |
and we have to contend with that idea with AI.
link |
But what's gonna motivate it is,
link |
I mean, I talk to people who really miss people who are gone
link |
and they would love to get something back,
link |
even if it isn't perfect.
link |
And that's what's gonna motivate this.
link |
And that person lives on in some form.
link |
And the more data we have,
link |
the more we're able to reconstruct that person
link |
and allow them to live on.
link |
And eventually as we go forward,
link |
we're gonna have more and more of this data
link |
because we're gonna have none of us
link |
that are inside our neocortex
link |
and we're gonna collect a lot of data.
link |
In fact, anything that's data is always collected.
link |
There is something a little bit sad,
link |
which is becoming, or maybe it's hopeful,
link |
which is more and more common these days,
link |
which when a person passes away,
link |
you have their Twitter account,
link |
and you have the last tweet they tweeted,
link |
like something they needed.
link |
And you can recreate them now
link |
with large language models and so on.
link |
I mean, you can create somebody that's just like them
link |
and can actually continue to communicate.
link |
I think that's really exciting
link |
because I think in some sense,
link |
like if I were to die today,
link |
in some sense I would continue on if I continued tweeting.
link |
I tweet, therefore I am.
link |
Yeah, well, I mean, that's one of the advantages
link |
of a replicant, they can recreate the communications
link |
Do you hope, do you think, do you hope
link |
humans will become a multi planetary species?
link |
You've talked about the phases, the six epochs,
link |
and one of them is reaching out into the stars in part.
link |
Yes, but the kind of attempts we're making now
link |
to go to other planetary objects
link |
doesn't excite me that much
link |
because it's not really advancing anything.
link |
It's not efficient enough?
link |
Yeah, and we're also putting out other human beings,
link |
which is a very inefficient way
link |
to explore these other objects.
link |
What I'm really talking about in the sixth epoch,
link |
the universe wakes up.
link |
It's where we can spread our super intelligence
link |
throughout the universe.
link |
And that doesn't mean sending a very soft,
link |
squishy creatures like humans.
link |
Yeah, the universe wakes up.
link |
I mean, we would send intelligence masses of nanobots
link |
which can then go out and colonize
link |
these other parts of the universe.
link |
Do you think there's intelligent alien civilizations
link |
out there that our bots might meet?
link |
Most people say yes, absolutely.
link |
I mean, and the universe is too big.
link |
And they'll cite the Drake equation.
link |
And I think in Singularity is Near,
link |
I have two analyses of the Drake equation,
link |
both with very reasonable assumptions.
link |
And one gives you thousands of advanced civilizations
link |
And another one gives you one civilization.
link |
And we know of one.
link |
A lot of the analyses are forgetting
link |
the exponential growth of computation.
link |
Because we've gone from where the fastest way
link |
I could send a message to somebody was with a pony,
link |
which was what, like a century and a half ago?
link |
To the advanced civilization we have today.
link |
And if you accept what I've said,
link |
go forward a few decades,
link |
you can have absolutely fantastic amount of civilization
link |
compared to a pony, and that's in a couple hundred years.
link |
Yeah, the speed and the scale of information transfer
link |
is growing exponentially in a blink of an eye.
link |
Now think about these other civilizations.
link |
They're gonna be spread out at cosmic times.
link |
So if something is like ahead of us or behind us,
link |
it could be ahead of us or behind us by maybe millions
link |
of years, which isn't that much.
link |
I mean, the world is billions of years old,
link |
14 billion or something.
link |
So even a thousand years, if two or 300 years is enough
link |
to go from a pony to fantastic amount of civilization,
link |
we would see that.
link |
So of other civilizations that have occurred,
link |
okay, some might be behind us, but some might be ahead of us.
link |
If they're ahead of us, they're ahead of us
link |
by thousands, millions of years,
link |
and they would be so far beyond us,
link |
they would be doing galaxy wide engineering.
link |
But we don't see anything doing galaxy wide engineering.
link |
So either they don't exist, or this very universe
link |
is a construction of an alien species.
link |
We're living inside a video game.
link |
Well, that's another explanation that yes,
link |
you've got some teenage kids in another civilization.
link |
Do you find compelling the simulation hypothesis
link |
as a thought experiment that we're living in a simulation?
link |
The universe is computational.
link |
So we are an example in a computational world.
link |
Therefore, it is a simulation.
link |
It doesn't necessarily mean an experiment
link |
by some high school kid in another world,
link |
but it nonetheless is taking place
link |
in a computational world.
link |
And everything that's going on
link |
is basically a form of computation.
link |
So you really have to define what you mean
link |
by this whole world being a simulation.
link |
Well, then it's the teenager that makes the video game.
link |
Us humans with our current limited cognitive capability
link |
have strived to understand ourselves
link |
and we have created religions.
link |
Whatever that is, do you think God exists?
link |
And if so, who is God?
link |
I alluded to this before.
link |
We started out with lots of particles going around
link |
and there's nothing that represents love and creativity.
link |
And somehow we've gotten into a world
link |
where love actually exists
link |
and that has to do actually with consciousness
link |
because you can't have love without consciousness.
link |
So to me, that's God, the fact that we have something
link |
where love, where you can be devoted to someone else
link |
and really feel the love, that's God.
link |
And if you look at the Old Testament,
link |
it was actually created by several different
link |
ravenants in there.
link |
And I think they've identified three of them.
link |
One of them dealt with God as a person
link |
that you can make deals with and he gets angry
link |
and he wrecks vengeance on various people.
link |
But two of them actually talk about God
link |
as a symbol of love and peace and harmony and so forth.
link |
That's how they describe God.
link |
So that's my view of God, not as a person in the sky
link |
that you can make deals with.
link |
It's whatever the magic that goes from basic elements
link |
to things like consciousness and love.
link |
Do you think one of the things I find
link |
extremely beautiful and powerful is cellular automata,
link |
which you also touch on?
link |
Do you think whatever the heck happens in cellular automata
link |
where interesting, complicated objects emerge,
link |
God is in there too?
link |
The emergence of love in this seemingly primitive universe?
link |
Well, that's the goal of creating a replicant
link |
is that they would love you and you would love them.
link |
There wouldn't be much point of doing it
link |
if that didn't happen.
link |
But all of it, I guess what I'm saying
link |
about cellular automata is it's primitive building blocks
link |
and they somehow create beautiful things.
link |
Is there some deep truth to that
link |
about how our universe works?
link |
Is the emergence from simple rules,
link |
beautiful, complex objects can emerge?
link |
Is that the thing that made us?
link |
Yeah, well. As we went through
link |
all the six phases of reality.
link |
That's a good way to look at it.
link |
It does make some point to the whole value
link |
of having a universe.
link |
Do you think about your own mortality?
link |
Are you afraid of it?
link |
Yes, but I keep going back to my idea
link |
of being able to expand human life quickly enough
link |
in advance of our getting there, longevity escape velocity,
link |
which we're not quite at yet,
link |
but I think we're actually pretty close,
link |
particularly with, for example, doing simulated biology.
link |
I think we can probably get there within,
link |
say, by the end of this decade, and that's my goal.
link |
Do you hope to achieve the longevity escape velocity?
link |
Do you hope to achieve immortality?
link |
Well, immortality is hard to say.
link |
I can't really come on your program saying I've done it.
link |
I've achieved immortality because it's never forever.
link |
A long time, a long time of living well.
link |
But we'd like to actually advance
link |
human life expectancy, advance my life expectancy
link |
more than a year every year,
link |
and I think we can get there within,
link |
by the end of this decade.
link |
How do you think we'd do it?
link |
So there's practical things in Transcend,
link |
the nine steps to living well forever, your book.
link |
You describe just that.
link |
There's practical things like health,
link |
exercise, all those things.
link |
Yeah, I mean, we live in a body
link |
that doesn't last forever.
link |
There's no reason why it can't, though,
link |
and we're discovering things, I think, that will extend it.
link |
But you do have to deal with,
link |
I mean, I've got various issues.
link |
Went to Mexico 40 years ago, developed salmonella.
link |
I created pancreatitis, which gave me
link |
a strange form of diabetes.
link |
It's not type one diabetes, because it's an autoimmune
link |
disorder that destroys your pancreas.
link |
I don't have that.
link |
But it's also not type two diabetes,
link |
because type two diabetes is your pancreas works fine,
link |
but your cells don't absorb the insulin well.
link |
I don't have that either.
link |
The pancreatitis I had partially damaged my pancreas,
link |
but it was a one time thing.
link |
It didn't continue, and I've learned now how to control it.
link |
But so that's just something that I had to do
link |
in order to continue to exist.
link |
Since your particular biological system,
link |
you had to figure out a few hacks,
link |
and the idea is that science would be able
link |
to do that much better, actually.
link |
Yeah, so I mean, I do spend a lot of time
link |
just tinkering with my own body to keep it going.
link |
So I do think I'll last till the end of this decade,
link |
and I think we'll achieve longevity, escape velocity.
link |
I think that we'll start with people
link |
who are very diligent about this.
link |
Eventually, it'll become sort of routine
link |
that people will be able to do it.
link |
So if you're talking about kids today,
link |
or even people in their 20s or 30s,
link |
that's really not a very serious problem.
link |
I have had some discussions with relatives
link |
who are like almost 100, and saying,
link |
well, we're working on it as quickly as possible.
link |
I don't know if that's gonna work.
link |
Is there a case, this is a difficult question,
link |
but is there a case to be made against living forever
link |
that a finite life, that mortality is a feature, not a bug,
link |
that living a shorter, so dying makes ice cream
link |
taste delicious, makes life intensely beautiful
link |
more than it otherwise might be?
link |
Most people believe that way, except if you present
link |
a death of anybody they care about or love,
link |
they find that extremely depressing.
link |
And I know people who feel that way
link |
20, 30, 40 years later, they still want them back.
link |
So I mean, death is not something to celebrate,
link |
but we've lived in a world where people just accept this.
link |
Life is short, you see it all the time on TV,
link |
oh, life's short, you have to take advantage of it
link |
and nobody accepts the fact that you could actually
link |
go beyond normal lifetimes.
link |
But anytime we talk about death or a death of a person,
link |
even one death is a terrible tragedy.
link |
If you have somebody that lives to 100 years old,
link |
we still love them in return.
link |
And there's no limitation to that.
link |
In fact, these kinds of trends are gonna provide
link |
greater and greater opportunity for everybody,
link |
even if we have more people.
link |
So let me ask about an alien species
link |
or a super intelligent AI 500 years from now
link |
that will look back and remember Ray Kurzweil version zero.
link |
Before the replicants spread,
link |
how do you hope they remember you
link |
in Hitchhiker's Guide to the Galaxy summary of Ray Kurzweil?
link |
What do you hope your legacy is?
link |
Well, I mean, I do hope to be around, so that's.
link |
Some version of you, yes.
link |
Do you think you'll be the same person around?
link |
I mean, am I the same person I was when I was 20 or 10?
link |
You would be the same person in that same way,
link |
but yes, we're different, we're different.
link |
All we have of that, all you have of that person
link |
is your memories, which are probably distorted in some way.
link |
Maybe you just remember the good parts,
link |
depending on your psyche.
link |
You might focus on the bad parts,
link |
might focus on the good parts.
link |
Right, but I mean, I still have a relationship
link |
to the way I was when I was earlier, when I was younger.
link |
How will you and the other super intelligent AIs
link |
remember you of today from 500 years ago?
link |
What do you hope to be remembered by this version of you
link |
before the singularity?
link |
Well, I think it's expressed well in my books,
link |
trying to create some new realities that people will accept.
link |
I mean, that's something that gives me great pleasure,
link |
and greater insight into what makes humans valuable.
link |
I'm not the only person who's tempted to comment on that.
link |
And optimism that permeates your work.
link |
Optimism about the future is ultimately that optimism
link |
paves the way for building a better future.
link |
Yeah, I agree with that.
link |
So you asked your dad about the meaning of life,
link |
and he said, love, let me ask you the same question.
link |
What's the meaning of life?
link |
This beautiful journey that we're on in phase four,
link |
reaching for phase five of this evolution
link |
and information processing, why?
link |
Well, I think I'd give the same answers as my father.
link |
Because if there were no love,
link |
and we didn't care about anybody,
link |
there'd be no point existing.
link |
Love is the meaning of life.
link |
The AI version of your dad had a good point.
link |
Well, I think that's a beautiful way to end it.
link |
Ray, thank you for your work.
link |
Thank you for being who you are.
link |
Thank you for dreaming about a beautiful future
link |
and creating it along the way.
link |
And thank you so much for spending
link |
your really valuable time with me today.
link |
It was my pleasure, and you have some great insights,
link |
both into me and into humanity as well, so I appreciate that.
link |
Thanks for listening to this conversation
link |
with Ray Kurzweil.
link |
To support this podcast,
link |
please check out our sponsors in the description.
link |
And now, let me leave you with some words
link |
from Isaac Asimov.
link |
It is change, continuous change, inevitable change
link |
that is the dominant factor in society today.
link |
No sensible decision can be made any longer
link |
without taking into account not only the world as it is,
link |
but the world as it will be.
link |
This, in turn, means that our statesmen,
link |
our businessmen, our everyman,
link |
must take on a science fictional way of thinking.
link |
Thank you for listening, and hope to see you next time.