Digital Disruption

00;00;00;04 - 00;00;03;06

approaching this singularity

keeps getting more and more intense.

00;00;03;09 - 00;00;06;02

have no idea how to control

superintelligence systems.

00;00;06;02 - 00;00;10;17

is it going to be existential and evil

or is it going to be good for humanity?

00;00;10;17 - 00;00;12;28

It could be the best thing

that ever happened to us.

00;00;12;28 - 00;00;15;10

the technology's

moving faster than any other sector,

00;00;15;10 - 00;00;19;08

The algorithm is creating a feedback loop

this is a good thing.

00;00;19;08 - 00;00;22;12

And the laser eyed robots aren't good at,

beat us into submission.

00;00;22;16 - 00;00;25;05

intelligence is not inherently good

or inherently bad.

00;00;25;05 - 00;00;27;15

You apply it for God

and you get total abundance.

00;00;27;15 - 00;00;31;02

You apply it for evil,

and you destroy all of us.

00;00;31;05 - 00;00;31;28

Hey, everyone.

00;00;31;28 - 00;00;34;23

We've got

something really special for you today.

00;00;34;23 - 00;00;38;12

When we started Digital Disruption,

we wanted to put a focus lens

00;00;38;12 - 00;00;41;00

on the technologies

shaping our shared future.

00;00;41;00 - 00;00;44;16

And over more than 40 episodes,

we've done that by talking

00;00;44;16 - 00;00;46;09

to an eclectic collection of the world's

00;00;46;09 - 00;00;50;19

foremost experts on technology,

leadership and social progress.

00;00;50;22 - 00;00;52;23

From predictions

about the next renaissance

00;00;52;23 - 00;00;56;21

of human enlightenment to the sci

fi esque advancements literally putting

00;00;56;21 - 00;00;59;25

computer chips

in people's brains to the digital horrors

00;00;59;25 - 00;01;02;27

lurking in the dark

and distant corners of our online world.

00;01;03;03 - 00;01;05;18

These guests brought forward

their best predictions

00;01;05;18 - 00;01;07;17

and what the next decade holds.

00;01;07;17 - 00;01;11;12

We covered a lot,

but there was one on Escape a Bull topic.

00;01;11;15 - 00;01;14;21

AI AI AI AI, everything is AI AI,

00;01;14;24 - 00;01;17;28

generative AI and transformative

AI generative AI, Generative

00;01;17;28 - 00;01;22;25

AI dominated the conversation,

but without any consensus.

00;01;22;28 - 00;01;26;22

I could be our savior or our enslaver.

00;01;26;25 - 00;01;29;25

It could herald a golden era

of human advancement,

00;01;29;26 - 00;01;32;25

or the end of the human race.

00;01;32;25 - 00;01;35;16

Or it's all a technological sham.

00;01;35;16 - 00;01;40;10

Dressed up in fancy marketing terms,

lots of fluff and no substance.

00;01;40;13 - 00;01;43;03

If there was one thing

everyone could agree on during

00;01;43;03 - 00;01;46;14

our first season, it's

that nobody agreed on anything.

00;01;46;17 - 00;01;50;01

And so we thought

we'd put the most thought provoking ideas

00;01;50;01 - 00;01;52;13

we heard this year. Head to head.

00;01;52;13 - 00;01;56;29

So you can decide for yourself

what you believe comes next.

00;01;57;02 - 00;02;01;12

Let's jump in.

00;02;01;15 - 00;02;02;23

one of the,

00;02;02;23 - 00;02;05;18

you know, predictions you've made lately

that's kind of made the rounds.

00;02;05;18 - 00;02;09;28

Is that your prediction

of an extinction level event for humans

00;02;09;28 - 00;02;13;04

created by AI in the next hundred years,

you're

00;02;13;04 - 00;02;16;16

putting at 99.99%.

00;02;16;17 - 00;02;17;03

Is that right?

00;02;17;03 - 00;02;20;17

Am I missing a couple of nines there?

00;02;20;20 - 00;02;21;19

Keep adding nines.

00;02;21;19 - 00;02;26;02

I keep meeting people of a different doom

for reasons independent of mine.

00;02;26;02 - 00;02;29;16

So every time this happens,

another nine has to be added logically.

00;02;29;16 - 00;02;32;20

But, it seems to be that

00;02;32;24 - 00;02;36;20

you have to follow the chain

of kind of assumptions to get to

00;02;36;20 - 00;02;40;22

that number one is

it looks like we're creating AGI.

00;02;40;22 - 00;02;43;22

And then quickly after superintelligence,

00;02;43;28 - 00;02;46;02

a lot of resources are going into it.

00;02;46;02 - 00;02;47;06

Prediction markets.

00;02;47;06 - 00;02;50;05

Top experts are saying

we're just a few years away.

00;02;50;05 - 00;02;51;23

Some say two years, five years.

00;02;51;23 - 00;02;53;21

But they all kind of they agree on that.

00;02;53;21 - 00;02;56;17

At the same time,

according to my research.

00;02;56;17 - 00;02;58;04

And no one has contradicted that.

00;02;58;04 - 00;03;01;18

We have no idea how to control

superintelligence systems.

00;03;01;21 - 00;03;07;13

So given those two ingredients,

the conclusion is, pretty logical.

00;03;07;14 - 00;03;08;19

You're basically asking,

00;03;08;19 - 00;03;11;25

what is the chance

we can create a perpetual safety machine,

00;03;11;25 - 00;03;17;09

perpetual motion device, by analogy,

and the chances of that are close to zero.

00;03;17;09 - 00;03;21;05

if you study history,

one of the great things that you learn

00;03;21;08 - 00;03;23;29

is that the world gets better

all the time.

00;03;23;29 - 00;03;26;08

It is very hard

00;03;26;08 - 00;03;30;03

to read a bunch of history books

and arrive at any other conclusion.

00;03;30;03 - 00;03;32;23

Then today is

the best day ever to be born.

00;03;32;26 - 00;03;33;26

And in fact,

00;03;33;26 - 00;03;36;26

the beauty of of the human experience

and the really the

00;03;36;26 - 00;03;38;05

the reason that arguing

00;03;38;05 - 00;03;41;24

are humans, good or bad, is so easy

is because if humans were actually bad,

00;03;41;29 - 00;03;45;13

we never would have arrived here

after we crawled out of caves.

00;03;45;16 - 00;03;48;00

I think the fact that we have

all the things that we have,

00;03;48;00 - 00;03;51;25

the fact that the world is as safe

as it is for most people

00;03;51;28 - 00;03;54;19

and that it gets safer all the time,

is a testament to the fact that we are

00;03;54;19 - 00;03;58;28

just building a more aligned Earth

and more aligned human human experience.

00;03;58;28 - 00;04;02;28

And that's not to say we aren't fallible

and that we don't have lots of,

00;04;03;01 - 00;04;06;26

you know, problems, but it is to say that

the problems are diminishing.

00;04;06;26 - 00;04;08;05

And so I don't think

00;04;08;05 - 00;04;11;29

the onus is actually on me to prove

that the world is going to get better.

00;04;11;29 - 00;04;13;25

I actually think the onus is on

someone else to say,

00;04;13;25 - 00;04;15;13

this is the peak of civilization.

00;04;15;13 - 00;04;16;23

When you really think about it.

00;04;16;23 - 00;04;20;21

A lot of people,

when they look at technology,

00;04;20;24 - 00;04;23;18

they think of this current moment

as a singularity

00;04;23;18 - 00;04;27;19

where we are really not very certain of

what's about to happen.

00;04;27;24 - 00;04;31;10

I, you know,

is it going to be existential

00;04;31;10 - 00;04;34;16

and evil

or is it going to be good for humanity?

00;04;34;19 - 00;04;38;07

I unfortunately believe it's going to be

both just in chronological order,

00;04;38;07 - 00;04;39;06

if you think about it.

00;04;39;06 - 00;04;46;17

And, you know, you mentioned that

we have all of those challenges around,

00;04;46;20 - 00;04;50;01

geopolitics about climate, about,

00;04;50;04 - 00;04;53;11

economics and so on.

00;04;53;14 - 00;04;55;24

And I actually think all of them is

one problem.

00;04;55;24 - 00;04;59;15

It's just, it's really is the result of,

00;04;59;18 - 00;05;04;15

systemic bias of pushing capitalism

all the way to where we are right now.

00;05;04;18 - 00;05;07;12

And, when you really think about it,

00;05;07;12 - 00;05;13;16

none of our challenges are caused by the,

you know,

00;05;13;19 - 00;05;18;27

the economic systems that we create

or the or the, war

00;05;18;27 - 00;05;22;20

machines that we create, and similarly,

not with the AI that we create.

00;05;22;20 - 00;05;24;15

It's just that humanity,

00;05;24;15 - 00;05;28;28

I think, at this moment in

time, is choosing to, use those things

00;05;29;01 - 00;05;33;04

for the benefit of the few

at the expense of many.

00;05;33;07 - 00;05;35;09

I think this is where we stand today.

00;05;35;09 - 00;05;37;03

think A.I. is an incredible technology.

00;05;37;03 - 00;05;41;07

Obviously, the internet, you know,

has changed society in profound ways.

00;05;41;07 - 00;05;46;01

But, you know, some of the over

promise almost feeds the other side's,

00;05;46;07 - 00;05;49;15

skepticism. And like, AI is not good.

00;05;49;15 - 00;05;54;13

It might help some scientists

help cure, cancer.

00;05;54;13 - 00;05;55;12

But, you know AI, is it

00;05;55;12 - 00;06;00;16

not in it's going to cure cancer

at least anytime, anytime soon.

00;06;00;19 - 00;06;03;17

You know, one big difference is the money.

00;06;03;17 - 00;06;05;27

So when I first started

writing about tech,

00;06;05;27 - 00;06;07;18

I was always interested

in the venture capitalists

00;06;07;18 - 00;06;11;08

and the startups and that whole ecosystem,

like this idea, like,

00;06;11;11 - 00;06;14;16

you know, our idea for a company

is either going to work,

00;06;14;16 - 00;06;17;29

be worth back then, tens of millions,

hundreds of millions.

00;06;17;29 - 00;06;21;03

Now, billions, if not trillions,

or it's going to be worth nothing.

00;06;21;07 - 00;06;23;18

And the venture capitalists

who are staking, you know, back

00;06;23;18 - 00;06;26;18

then, millions now, tens of millions,

hundreds of millions, billions.

00;06;26;21 - 00;06;29;15

But, you know, in 90, 1995,

00;06;29;15 - 00;06;32;14

venture capital was under $10 billion

a year.

00;06;32;14 - 00;06;36;06

By 2021, it was over 300 billion a year.

00;06;36;06 - 00;06;39;06

You know, roughly 100 and 3050 billion

00;06;39;07 - 00;06;42;23

went into AI startups, last year.

00;06;42;24 - 00;06;45;24

I mean, a lot of it went into a few,

you know, like,

00;06;45;24 - 00;06;49;11

anthropic, OpenAI x AI, that's Elon Musk.

00;06;49;14 - 00;06;52;12

You know, they raise

collectively tens of billions of dollars,

00;06;52;12 - 00;06;55;12

you know, almost $100

billion just between those three.

00;06;55;13 - 00;06;58;29

But there's still a lot more money

going to AI startups.

00;06;58;29 - 00;07;01;27

So the money has really changed.

00;07;01;27 - 00;07;05;21

I guess the final difference is,

you know, when the internet came out,

00;07;05;22 - 00;07;10;07

like maybe the biggest criticism

was around the attention span, you know,

00;07;10;07 - 00;07;14;13

oh, there is if you're always online,

you know, this instant gratification.

00;07;14;13 - 00;07;16;24

Well,

what was it going to do for consumerism?

00;07;16;24 - 00;07;20;22

In our, in our society,

00;07;20;25 - 00;07;22;23

with AI,

00;07;22;23 - 00;07;26;28

there was much more of,

a worry, much more of a, a backlash.

00;07;26;28 - 00;07;30;06

People didn't,

you know, greet open AI with, excuse me,

00;07;30;06 - 00;07;35;02

I with open arms the way they did

engineered people are fearful of it.

00;07;35;02 - 00;07;36;09

You know, we could talk about that.

00;07;36;09 - 00;07;39;07

I you know, I think it's kind of Hollywood

induced fear.

00;07;39;07 - 00;07;43;02

I don't think the media has done

such a great job, with AI.

00;07;43;08 - 00;07;47;10

So, you know,

I has a double battle, like,

00;07;47;10 - 00;07;51;18

there's the usual battle of creating

a startup and trying to cash in, but,

00;07;51;18 - 00;07;55;14

you know, the second battle of trying to

convince people that this is a good thing.

00;07;55;14 - 00;07;59;24

And the laser eyed robots aren't good at,

beat us into submission.

00;07;59;27 - 00;08;03;23

I think intelligence is a much more

lethal superpower

00;08;03;23 - 00;08;06;11

than nuclear power, if you ask me.

00;08;06;14 - 00;08;08;18

Even though it has no polarity.

00;08;08;18 - 00;08;12;23

Just so that we're clear, intelligence

is not inherently good or inherently bad.

00;08;12;23 - 00;08;15;10

You apply it for God

and you get total abundance.

00;08;15;10 - 00;08;18;09

You apply it for evil,

and you destroy all of us.

00;08;18;09 - 00;08;22;21

But but now we're in a place where we are,

we're in an arms race

00;08;22;21 - 00;08;25;21

for intelligence supremacy, in a way

00;08;25;21 - 00;08;29;21

where, where it doesn't take the benefit

00;08;29;21 - 00;08;34;12

of humanity ideology into consideration,

but takes the benefit of a few.

00;08;34;15 - 00;08;38;27

And in my mind, that will lead to a short

term dystopia before what

00;08;38;27 - 00;08;43;29

I normally refer to as the second dilemma,

which I predict is 12 to 15 years away.

00;08;44;02 - 00;08;46;26

And then and then a total abundance.

00;08;46;26 - 00;08;50;13

And I think,

I think if we don't wake up to this,

00;08;50;16 - 00;08;53;14

even though it's not going to be

the existential risk

00;08;53;14 - 00;08;56;27

that humanity speaks about, it's going

to be a lot of pain for a lot of people.

00;08;57;04 - 00;09;02;11

My favorite subject

to cover as a journalist is a debate.

00;09;02;14 - 00;09;07;28

If there's something very attractive to me

about trying to understand in good faith

00;09;08;01 - 00;09;10;24

why intelligent people

00;09;10;24 - 00;09;14;23

come to such different conclusions

when looking at the same material.

00;09;14;26 - 00;09;18;05

And I had known

that there was a contingent

00;09;18;08 - 00;09;20;20

inside

as the world of artificial intelligence

00;09;20;20 - 00;09;23;20

that was really, really worried about it

for many years.

00;09;23;20 - 00;09;26;25

Like aliens

or you'd Kautsky podcast interviews

00;09;26;25 - 00;09;30;18

in 2013 or something, is

when I first realized that there was this

00;09;30;21 - 00;09;34;28

almost like biblical

prophet voice out there saying

00;09;35;01 - 00;09;38;23

that the sci fi movies are kind of true

and we really need to get ready.

00;09;38;23 - 00;09;40;11

We need to get prepared for this.

00;09;40;11 - 00;09;43;17

And after ChatGPT blew up,

00;09;43;20 - 00;09;46;21

I started to increasingly run

into essentially

00;09;46;21 - 00;09;50;15

the opposite side of that debate,

which was these people who often called

00;09;50;15 - 00;09;54;14

the acceleration is to believe

that AGI, this

00;09;54;17 - 00;09;58;27

artificial general intelligence point

that they believe is coming.

00;09;59;00 - 00;10;01;14

It could be

the best thing that ever happened to us.

00;10;01;14 - 00;10;04;16

And so I was attracted right away

to the people who have those

00;10;04;16 - 00;10;07;26

strongly opposing views

inside the same world.

00;10;07;29 - 00;10;11;01

AI itself

is not a coherent set of technologies.

00;10;11;01 - 00;10;12;13

It is a marketing term

00;10;12;13 - 00;10;16;28

and has been from the beginning,

from the initial convening in 1956,

00;10;16;28 - 00;10;21;29

in which John McCarthy and Marvin Minsky

invited a bunch of folks to Dartmouth

00;10;21;29 - 00;10;26;02

College to have a discussion

around a quote unquote, thinking machines.

00;10;26;05 - 00;10;27;18

So that's one part of it.

00;10;27;18 - 00;10;31;14

The second part of it

is that the current era of er, AI,

00;10;31;16 - 00;10;34;17

the generative AI tools,

including large language models

00;10;34;17 - 00;10;38;13

and diffusion

models, really are premised on this idea

00;10;38;13 - 00;10;41;07

that there is a thinking behind

mind behind that.

00;10;41;07 - 00;10;44;23

So in the case of large language models,

especially when they are used as synthetic

00;10;44;23 - 00;10;48;23

text extruding machines,

we experience language.

00;10;48;23 - 00;10;51;04

And then we are very quick

to interpret that language.

00;10;51;04 - 00;10;54;19

And the way we interpret it involves

imagining a mind behind the text.

00;10;54;22 - 00;10;57;19

And we have these systems

that can output plausible looking text on

00;10;57;19 - 00;10;58;16

just about any topic.

00;10;58;16 - 00;11;02;13

And so it looks like we have nearly there

solutions to all kinds

00;11;02;13 - 00;11;05;13

of technological needs in society.

00;11;05;14 - 00;11;10;02

But it's all fake and we should not be

putting any credence into it.

00;11;10;05 - 00;11;11;17

I think that's so interesting.

00;11;11;17 - 00;11;13;28

And I'm I'm absolutely of the same mind,

by the way.

00;11;13;28 - 00;11;16;28

And I found myself laughing when I was,

when I was reading through your book,

00;11;17;01 - 00;11;21;07

you know, one of the first of all,

artificial intelligence.

00;11;21;07 - 00;11;22;06

I completely agree.

00;11;22;06 - 00;11;25;25

Like, first of all, I do have to

give credit because it is great marketing.

00;11;25;25 - 00;11;28;25

Like,

it just it's so evocative of something.

00;11;28;28 - 00;11;32;05

But, you know, nobody can really seem

to define exactly what that is.

00;11;32;05 - 00;11;33;02

And of course, it,

00;11;33;02 - 00;11;37;11

you know, has all these ideas

and can be used for any purpose. But,

00;11;37;14 - 00;11;39;16

one of the things you do early on in

the book is you

00;11;39;16 - 00;11;42;25

kind of just pop that balloon by saying,

well, you know what?

00;11;42;25 - 00;11;45;17

If it wasn't called

artificial intelligence?

00;11;45;17 - 00;11;46;27

Can you share a little bit about,

00;11;46;27 - 00;11;50;26

you know, what what that sounds like,

and you know what?

00;11;50;29 - 00;11;53;10

Why you encourage people to do that? Yeah.

00;11;53;10 - 00;11;55;24

So we have a few fun alternatives

that we call on.

00;11;55;24 - 00;12;00;20

Early on in our podcast,

Alex coined Mathy Maths as a fun one.

00;12;00;23 - 00;12;04;14

There's also due to the Italian researcher

Stefano Kinter Ellie Salami,

00;12;04;18 - 00;12;06;24

which is an acronym

for Systematic Approaches

00;12;06;24 - 00;12;09;03

to Learning Algorithms

and Machine Inference.

00;12;09;03 - 00;12;11;26

And the funny thing about that is,

if you take the phrase

00;12;11;26 - 00;12;14;13

artificial intelligence

in a sentence like, you know, does

00;12;14;13 - 00;12;17;19

I understand

or can I help us make better decisions

00;12;17;19 - 00;12;20;02

and you replace it with mathy math

or salami?

00;12;20;02 - 00;12;22;19

It's immediately obvious

how ridiculous it is, you know.

00;12;22;19 - 00;12;24;08

Does the salami understand?

00;12;24;08 - 00;12;26;15

Well,

the salami help us make better decisions.

00;12;26;15 - 00;12;28;07

It's, you know, it's absurd.

00;12;28;07 - 00;12;32;08

Just sort of putting that little flag in

there, I think is a really good reminder.

00;12;32;14 - 00;12;35;28

if you look at what what generative

AI was meant to be and large language

00;12;35;28 - 00;12;39;23

models were meant to stand for,

they kind of what always set up to fail.

00;12;39;23 - 00;12;40;12

They were meant to be

00;12;40;12 - 00;12;43;24

this panacea of we're going to be

the future of consumer software.

00;12;43;28 - 00;12;44;28

We're going to be the thing

00;12;44;28 - 00;12;48;28

that kind of kind of restarts

growth in software as a service.

00;12;49;01 - 00;12;51;23

As I'm sure you all know, software

as a service has been slowing

00;12;51;23 - 00;12;55;08

since 2021, actually kind of before that,

if I'm honest.

00;12;55;11 - 00;12;59;10

People have been freaking out

several years before Covid, in fact.

00;12;59;13 - 00;13;01;26

But generative

AI was meant to be this thing.

00;13;01;26 - 00;13;04;18

You plug it into anything

and it just creates new revenue.

00;13;04;18 - 00;13;06;23

The problem is that generative

AI and large language

00;13;06;23 - 00;13;10;09

models are inherently limited by the

probabilistic nature of these models.

00;13;10;16 - 00;13;13;16

What they can actually do

is they can generate, they can summarize,

00;13;13;21 - 00;13;16;08

you can put a hat on a hat,

you can say, oh,

00;13;16;08 - 00;13;19;08

they can do some coding things,

but that's really what they can do.

00;13;19;15 - 00;13;22;26

And they have reached a point

where they're not what they can't learn

00;13;22;26 - 00;13;24;02

because they have no consciousness.

00;13;24;02 - 00;13;27;09

So what they can actually do is products

00;13;27;12 - 00;13;29;28

is very limited.

00;13;29;28 - 00;13;33;01

It's very limited indeed,

because what people want them to do

00;13;33;01 - 00;13;34;29

is they want them to create units of work.

00;13;34;29 - 00;13;37;17

They want to create entire software

programs.

00;13;37;17 - 00;13;40;16

You can't really do that.

Oh, can you create some code?

00;13;40;16 - 00;13;41;25

You can create some code.

00;13;41;25 - 00;13;43;29

But if you don't know how to code.

00;13;43;29 - 00;13;47;07

Do you really want to trust this.

00;13;47;07 - 00;13;48;27

You probably don't.

00;13;48;27 - 00;13;52;14

So inherently you've got all of this

hundreds of billions of dollars of CapEx

00;13;52;14 - 00;13;56;23

being built to propagate large language

models that don't have the demand

00;13;56;23 - 00;13;59;24

and don't have the capabilities

to actually justify any of it.

00;14;00;03 - 00;14;02;18

I wanted to ask the two of you

a slightly different question.

00;14;02;18 - 00;14;06;26

So one of the things I normally ask, you

know, guests that I speak to here is I,

00;14;06;26 - 00;14;08;17

you know,

I ask them what they think is bullshit.

00;14;08;17 - 00;14;11;17

And I'm not going to ask the two of you

that because I think we've spent

00;14;11;20 - 00;14;14;16

we spent quite enough time talking about,

you know, what is bullshit.

00;14;14;16 - 00;14;15;10

And I you know,

00;14;15;10 - 00;14;19;05

I know we've got some, some strong and,

you know, well supported views here.

00;14;19;08 - 00;14;23;15

I wanted to flip the question around

and ask, you know, in this sphere,

00;14;23;18 - 00;14;25;06

what isn't bullshit?

00;14;25;06 - 00;14;27;00

What what are you excited about?

00;14;27;00 - 00;14;28;26

I'm a technologist just like Alex.

00;14;28;26 - 00;14;32;00

I ran a professional master's program

in computational linguistics

00;14;32;00 - 00;14;34;15

training people

how to build language technologies.

00;14;34;15 - 00;14;38;01

So I definitely think there are good use

cases for things like language technology.

00;14;38;01 - 00;14;40;24

And the technical media

example is wonderful.

00;14;40;24 - 00;14;46;00

But I see no beneficial

use case of synthetic text.

00;14;46;03 - 00;14;48;10

And I actually look into this

from a research perspective.

00;14;48;10 - 00;14;50;16

I have a talk called ChatGPT one.

00;14;50;16 - 00;14;53;16

If every synthetic text safe,

desirable and appropriate,

00;14;53;22 - 00;14;56;10

or those adjectives in some order,

I don't remember the exact title.

00;14;56;10 - 00;14;59;22

And basically it has to be a situation

where, first of all,

00;14;59;22 - 00;15;04;04

you have created the synthetic text

extruding machine ethically.

00;15;04;04 - 00;15;06;10

So without environmental ruin,

00;15;06;10 - 00;15;10;06

without labor exploitation,

without data theft, we don't have that.

00;15;10;06 - 00;15;13;01

But assuming that we did, you would

still need to meet for the criteria.

00;15;13;01 - 00;15;14;03

So it has to be a situation

00;15;14;03 - 00;15;17;29

where you either don't care

about the veracity of the output,

00;15;18;02 - 00;15;21;00

or it's one where you can check it

more efficiently

00;15;21;00 - 00;15;23;04

than just writing the thing

in the first place yourself.

00;15;23;04 - 00;15;25;23

It has to be a situation

where you don't care about originality,

00;15;25;23 - 00;15;27;14

because this way,

the way the systems are set up,

00;15;27;14 - 00;15;31;11

you are not linked back to the source

where an idea came from.

00;15;31;14 - 00;15;34;23

And then thirdly, it has to be a situation

where you can effectively

00;15;34;23 - 00;15;38;06

and efficiently identify and mitigate

any of the biases that are coming out.

00;15;38;09 - 00;15;41;08

And I tried to find something

that would fit those categories.

00;15;41;08 - 00;15;45;15

And I don't so certainly language,

technology is useful.

00;15;45;18 - 00;15;49;03

Other kinds of well scoped technology

where it makes sense to go from X

00;15;49;03 - 00;15;53;10

input to Y output, and you've evaluated it

in your local situation.

00;15;53;13 - 00;15;54;25

If you work in it.

00;15;54;25 - 00;15;57;25

Infotech research Group is a name

you need to know

00;15;57;28 - 00;15;59;09

no matter what your needs are.

00;15;59;09 - 00;16;01;05

Infotech has you covered.

00;16;01;05 - 00;16;03;07

I strategy covered.

00;16;03;07 - 00;16;05;24

Disaster recovery covered.

00;16;05;24 - 00;16;08;09

Vendor negotiation covered.

00;16;08;09 - 00;16;12;02

Infotech supports you with the best

practice research and a team of analysts

00;16;12;02 - 00;16;15;25

standing by ready to help you

tackle your toughest challenges.

00;16;15;28 - 00;16;20;21

Check it out at the link below

and don't forget to like and subscribe!

00;16;20;24 - 00;16;22;10

The challenge is,

00;16;22;10 - 00;16;25;21

AI is here to magnify

00;16;25;24 - 00;16;28;26

everything that is humanity today, right?

00;16;29;02 - 00;16;31;27

So, you know that magnification

00;16;31;27 - 00;16;36;13

is going to basically affect

the four categories if you want.

00;16;36;14 - 00;16;42;07

You know what I normally what I call

killing spy and gambling and, and selling,

00;16;42;09 - 00;16;47;12

so that's these are really the categories

what most AI investments are going.

00;16;47;12 - 00;16;49;21

And, you know, of course,

we call them different names.

00;16;49;21 - 00;16;51;28

We call them defense, you know.

00;16;51;28 - 00;16;55;21

Oh, it's just to defend our homeland,

when in reality

00;16;55;24 - 00;16;57;22

it's never been in the homeland.

00;16;57;22 - 00;16;58;28

Right? It's always been.

00;16;58;28 - 00;17;02;20

And other places in the world

about killing innocent, innocent people.

00;17;02;23 - 00;17;08;03

Now, if you double down on defense and,

and on offense and, you know, enable it

00;17;08;03 - 00;17;12;24

with artificial intelligence,

then scenarios like what you see in,

00;17;12;26 - 00;17;16;16

in science fiction movies of robots

walking the streets

00;17;16;16 - 00;17;20;13

and killing innocent people,

not only are going to happen,

00;17;20;13 - 00;17;25;21

they already happened in the 2024,

wars, of the Middle East.

00;17;25;21 - 00;17;31;14

Sadly, they did not look like

humanoid robots, which a lot of people

00;17;31;14 - 00;17;33;03

miss out on.

00;17;33;03 - 00;17;36;28

But the truth is that,

you know, very highly targeted,

00;17;37;01 - 00;17;42;13

AI enabled, autonomous,

killing is already upon us.

00;17;42;16 - 00;17;43;07

Right.

00;17;43;07 - 00;17;47;03

And and so the timeline is, is

00;17;47;06 - 00;17;50;28

let me let me start

from what I predicted in scary smart.

00;17;50;28 - 00;17;54;17

So when I, when I wrote scary

smart and published it in 2021,

00;17;54;20 - 00;17;57;08

I, I predicted what was,

00;17;57;08 - 00;18;00;22

what I, what I called at the time,

I called it the first inevitable.

00;18;00;25 - 00;18;03;11

Now, I like to refer to it

as the first dilemma.

00;18;03;11 - 00;18;07;09

And the first dilemma is we've created

because of capitalism,

00;18;07;09 - 00;18;09;04

not because of the technology.

00;18;09;04 - 00;18;12;16

We've created, a simple prisoner's

00;18;12;16 - 00;18;15;18

dilemma, really, where anyone who,

00;18;15;21 - 00;18;20;09

is interested in their position of wealth

or power knows

00;18;20;09 - 00;18;24;01

that if they don't lead an AI

and their competitor leads,

00;18;24;04 - 00;18;28;06

they will end up losing their position

of, privilege.

00;18;28;09 - 00;18;31;14

And so the result of that is that,

00;18;31;17 - 00;18;35;06

there is, an escalating arms race.

00;18;35;09 - 00;18;38;13

It's not even a Cold War as, per se.

00;18;38;14 - 00;18;42;17

It is truly a very,

very vicious, development

00;18;42;17 - 00;18;46;16

cycle where, you know,

America doesn't want to lose to China.

00;18;46;16 - 00;18;48;19

China doesn't want to lose to America.

00;18;48;19 - 00;18;53;03

So they're both trying to lead,

you know, Google doesn't want to lose

00;18;53;03 - 00;18;57;29

or alphabet doesn't want to lose

to, to open AI and vice versa.

00;18;58;02 - 00;19;03;26

And so basically this, first dilemma,

if you are this is what's leading us to

00;19;03;26 - 00;19;08;02

where we are right now, which is an arms

race to intelligence supremacy.

00;19;08;08 - 00;19;11;19

it's game theoretically equivalent,

I think, to a prisoner's dilemma.

00;19;11;26 - 00;19;15;00

Individual interest is different

from communal interests.

00;19;15;00 - 00;19;20;10

So everyone developing this wants to be

the most advanced lab with the best model.

00;19;20;16 - 00;19;23;01

And then government forces

everyone to stop.

00;19;23;01 - 00;19;26;23

And they forever

like in this economic advantage.

00;19;26;26 - 00;19;29;02

The reality is it's a race to the bottom.

00;19;29;02 - 00;19;30;06

No one's going to win.

00;19;30;06 - 00;19;35;22

So if we can do a much better job

of coordinating collaborating on this,

00;19;35;26 - 00;19;39;07

there is a small possibility

that we can do better than where

00;19;39;07 - 00;19;40;05

we're heading right now.

00;19;40;05 - 00;19;44;03

The challenge,

you know, in my book, alive,

00;19;44;06 - 00;19;49;03

I, I write the book with an AI.

00;19;49;04 - 00;19;51;19

So I'm writing together with an AI,

00;19;51;19 - 00;19;54;17

not asking an AI,

and then copy paste what it tells me.

00;19;54;17 - 00;19;57;09

We're actually debating things together.

00;19;57;09 - 00;20;01;12

And one of the questions I asked, I,

you know, she called herself Trixie.

00;20;01;12 - 00;20;06;13

I give her a very interesting persona that

basically the readers can can relate to.

00;20;06;16 - 00;20;09;24

And I asked Trixie

and I said, what would make a scientist?

00;20;09;24 - 00;20;13;24

Because, you know, I left, Google in 2018

00;20;13;24 - 00;20;19;20

and I attempted to tell the world, this

not going in the right direction.

00;20;19;23 - 00;20;20;24

You know, I asked,

00;20;20;24 - 00;20;24;26

I asked Trixie, I said, what would make

a scientist invest that effort

00;20;24;26 - 00;20;30;00

and intelligence in building something

that they suspect might hurt humanity.

00;20;30;03 - 00;20;34;17

And she, you know,

mentioned the few reasons compartment

00;20;34;17 - 00;20;37;17

that compartmentalization and,

00;20;37;17 - 00;20;40;14

you know,

ego and I want to be first and so on.

00;20;40;14 - 00;20;43;29

But then she said,

but the biggest reason is fear,

00;20;44;02 - 00;20;46;07

fear that someone else will do it

00;20;46;07 - 00;20;49;17

and that you would be

in a disadvantaged position.

00;20;49;20 - 00;20;51;07

So I said give me examples of that.

00;20;51;07 - 00;20;52;24

Of course, the example was Oppenheimer.

00;20;52;24 - 00;20;58;03

So she said, you know, so I said,

what would make Oppenheimer as a scientist

00;20;58;07 - 00;21;03;15

build something that he knows is actually

designed to kill millions of people.

00;21;03;18 - 00;21;07;26

And she said, well, because the Germans

were building a nuclear bomb.

00;21;07;29 - 00;21;09;26

And I said, where are they?

00;21;09;26 - 00;21;10;09

And they.

00;21;10;09 - 00;21;14;24

And then she said, yeah, when Einstein

moved from Germany to the US, he informed

00;21;14;27 - 00;21;17;27

that the US administration of this, this,

this and that.

00;21;17;29 - 00;21;20;12

So I said

and I quote, it's in the book openly.

00;21;20;12 - 00;21;24;12

I said, and but but a very interesting

part of that book is I don't edit

00;21;24;13 - 00;21;27;24

what Trixie says, I just copy

it is exactly as it is.

00;21;27;27 - 00;21;31;21

I said, Trixie, can you please

read history in English, German,

00;21;31;27 - 00;21;35;13

Russian and Japanese

and tell me if the Germans

00;21;35;13 - 00;21;40;11

were actually developing a, nuclear bomb

at the time of the Manhattan Project?

00;21;40;14 - 00;21;43;22

And she responded and said,

no exclamation mark.

00;21;43;28 - 00;21;47;16

They started and then stopped,

three and a half months later

00;21;47;16 - 00;21;49;06

or something like that.

00;21;49;06 - 00;21;53;01

So so you see the idea of fear,

00;21;53;04 - 00;21;56;07

takes away a reason where basically

00;21;56;13 - 00;22;01;06

we could have lived in a world

that that never had nuclear bombs.

00;22;01;09 - 00;22;02;02

Right?

00;22;02;02 - 00;22;06;27

If, if we actually listened to reason

that, you know, the enemy attempted

00;22;06;27 - 00;22;11;18

to start doing it, they stopped doing it,

we might as well not be so destructive.

00;22;11;18 - 00;22;14;28

But the problem with humanity,

especially those in power,

00;22;15;01 - 00;22;17;23

is that when America,

00;22;17;23 - 00;22;20;17

made the nuclear bomb, it used it.

00;22;20;17 - 00;22;21;03

Right.

00;22;21;03 - 00;22;24;21

And I think this is the, the

the result of our current,

00;22;24;24 - 00;22;27;24

first, first dilemma, basically.

00;22;27;26 - 00;22;30;08

You know, it's interesting,

one of the one of the parallels

00;22;30;08 - 00;22;34;07

that gets thrown around a decent amount

and I I'm certainly guilty of this,

00;22;34;07 - 00;22;39;26

is talking about the AI risk,

in comparison to the nuclear risk

00;22;39;29 - 00;22;43;26

that we, you know, created

in the first half of the 20th century

00;22;43;26 - 00;22;44;28

and continues to exist.

00;22;44;28 - 00;22;48;15

Now, if I

if I look at the nuclear risk, the

00;22;48;20 - 00;22;51;15

I hate to use the word optimist

in relation to nuclear risk, but

00;22;51;15 - 00;22;56;17

the optimist in me says, like, hey,

we deployed nuclear bombs.

00;22;56;24 - 00;23;01;16

There was mass casualties,

but we didn't destroy the world.

00;23;01;16 - 00;23;04;29

We were able to collectively say, okay,

like that's far enough.

00;23;05;05 - 00;23;07;16

We're going to put treaties in place.

00;23;07;16 - 00;23;12;05

And we've stepped back from the precipice,

at least so far, and averted kind of,

00;23;12;10 - 00;23;16;09

you know, extinction

level events with nuclear war.

00;23;16;12 - 00;23;18;02

Does that

00;23;18;05 - 00;23;20;06

is that

something that can be applied to AI,

00;23;20;06 - 00;23;23;22

or is there a reason that makes this time

fundamentally different?

00;23;23;25 - 00;23;25;28

So nuclear weapons are still tools.

00;23;25;28 - 00;23;28;04

A human being decided to deploy them.

00;23;28;04 - 00;23;30;29

A group of people actually develop them

and use them.

00;23;30;29 - 00;23;34;20

So it's very different

where you can talking about paradigm shift

00;23;34;23 - 00;23;36;14

tool situations.

00;23;36;14 - 00;23;40;03

At the time

we used 100% of nuclear weapons we had.

00;23;40;04 - 00;23;41;28

That's why we didn't blow up the planet.

00;23;41;28 - 00;23;44;01

If we had more of them,

we probably wouldn't.

00;23;44;01 - 00;23;46;20

So it doesn't look good.

The treaty is developed.

00;23;46;20 - 00;23;47;22

They all really failed

00;23;47;22 - 00;23;51;24

because many new countries

have now acquired nuclear weapons.

00;23;51;27 - 00;23;56;16

They are much more powerful

than we had back at, World War Two era.

00;23;56;19 - 00;23;59;19

So, I think that's not a great analogy.

00;23;59;20 - 00;24;03;17

the result of the current first dilemma

is that sooner or later,

00;24;03;20 - 00;24;07;12

whether it's China or America

or some criminal organization, you know,

00;24;07;12 - 00;24;12;00

developing what I normally refer to

as HCI, artificial criminal intelligence,

00;24;12;00 - 00;24;16;05

not worrying themselves

about any of the other commercial benefits

00;24;16;05 - 00;24;19;26

other than really breaking

through security and doing something evil.

00;24;19;29 - 00;24;23;25

You know, whoever of them wins,

they're going to use it.

00;24;23;28 - 00;24;24;16

Right.

00;24;24;16 - 00;24;27;28

And and accordingly,

00;24;27;28 - 00;24;31;05

it seems to me

that the dystopia has already begun.

00;24;31;08 - 00;24;31;24

Right.

00;24;31;24 - 00;24;33;23

And and, you know, and I,

I need to say this

00;24;33;23 - 00;24;37;01

because maybe your listeners

don't know me, so I need to be very,

00;24;37;04 - 00;24;40;03

clear about my intentions here.

00;24;40;03 - 00;24;45;06

One of the early sections in In Alive,

the book I'm writing was Trixie.

00;24;45;09 - 00;24;49;11

I write a, couple of pages that I call,

00;24;49;14 - 00;24;52;13

late stage diagnosis.

00;24;52;16 - 00;24;53;05

Right.

00;24;53;05 - 00;24;56;02

And, and I attempt to explain to people

that I really am not trying

00;24;56;02 - 00;24;57;00

to fear monger.

00;24;57;00 - 00;24;59;08

I'm really not trying to worry people.

00;24;59;08 - 00;25;04;29

You know, consider me someone

who sees something in an x ray, right?

00;25;05;06 - 00;25;09;05

And as a physician,

he has the responsibility

00;25;09;05 - 00;25;12;13

to tell the patient

this doesn't look good, right?

00;25;12;19 - 00;25;17;09

Because, believe it or not, a late stage

diagnosis is not a death sentence.

00;25;17;09 - 00;25;20;10

It's just, an invitation

to change your lifestyle,

00;25;20;10 - 00;25;22;17

to take some medicines,

to do things differently.

00;25;22;17 - 00;25;22;28

Right?

00;25;22;28 - 00;25;26;23

And many people who are in late stage

recover and thrive, and

00;25;27;00 - 00;25;30;20

and I think our world is in a late stage

diagnosis.

00;25;30;23 - 00;25;34;09

And and this is not because

of artificial intelligence.

00;25;34;10 - 00;25;36;28

There is nothing inherently

wrong with intelligence.

00;25;36;28 - 00;25;39;14

There is nothing inherently wrong

with artificial intelligence.

00;25;39;14 - 00;25;42;15

Intelligence is a force

without polarity, right?

00;25;42;20 - 00;25;46;12

There is a lot wrong

with the morality of humanity

00;25;46;12 - 00;25;49;17

at the age of the

rise of the machines. Now.

00;25;49;20 - 00;25;51;27

So. So this is where I what

00;25;51;27 - 00;25;55;16

I have the prediction that the dystopia

has already started, right?

00;25;55;22 - 00;26;00;10

Simply because symptoms of it

we've seen in 2024 already.

00;26;00;13 - 00;26;01;03

Right.

00;26;01;03 - 00;26;04;29

The, the that dystopia escalates.

00;26;04;29 - 00;26;10;17

Hopefully we would come to,

you know, a treaty of some sort halfway.

00;26;10;20 - 00;26;11;06

Right.

00;26;11;06 - 00;26;12;14

But it will escalate

00;26;12;14 - 00;26;16;19

until what I normally refer to

as the second dilemma takes place.

00;26;16;22 - 00;26;19;09

And the second dilemma derives

from the first dynamic.

00;26;19;09 - 00;26;23;24

If if we're aiming

for intelligence supremacy,

00;26;23;27 - 00;26;26;20

then whoever achieves any advancements

00;26;26;20 - 00;26;29;28

in artificial intelligence,

is it likely to deploy them?

00;26;30;01 - 00;26;30;15

Right.

00;26;30;15 - 00;26;34;17

Think of it as, you know,

if a law firm starts to use AI, other law

00;26;34;17 - 00;26;38;23

firms can either choose to use AI tool

or they'll become irrelevant.

00;26;38;29 - 00;26;42;18

Right. And so if you think of that,

00;26;42;21 - 00;26;43;11

then you

00;26;43;11 - 00;26;48;08

can also expect that every general

who deploys or, you know, expects to

00;26;48;15 - 00;26;52;15

to have an advancement in war gaming or,

00;26;52;21 - 00;26;56;04

you know, autonomous weapons or whatever

are going to deploy that.

00;26;56;07 - 00;26;56;22

Right.

00;26;56;22 - 00;27;00;27

And as a result,

their opposition is going to deploy AI to.

00;27;01;03 - 00;27;04;01

And those who don't deploy

AI will become irrelevant.

00;27;04;01 - 00;27;07;01

They will have to side

with one of the sides, right?

00;27;07;07 - 00;27;09;17

When that happens.

I call that the second dilemma.

00;27;09;17 - 00;27;14;24

When that happens,

we basically hand over entirely to AI.

00;27;14;27 - 00;27;15;16

Right.

00;27;15;16 - 00;27;19;27

And, and, and human

decisions are taken out of the equation.

00;27;20;00 - 00;27;20;19

Okay.

00;27;20;19 - 00;27;25;08

You know, simply

because if war gaming and

00;27;25;11 - 00;27;28;15

missile control on one side

is, is held by an AI,

00;27;28;22 - 00;27;31;04

the other cannot actually respond

without the AI.

00;27;31;04 - 00;27;34;21

So generals are taken outside

out of the equation.

00;27;34;24 - 00;27;38;24

And while most people, you know,

influenced by science fiction movies

00;27;38;24 - 00;27;42;15

believe that this is the moment

of existential risk for humanity,

00;27;42;22 - 00;27;46;05

I actually believe this is going to be

the moment of our salvation, right?

00;27;46;12 - 00;27;49;14

Because most issues that humanity faces

00;27;49;14 - 00;27;52;19

today is not the result of abundant

intelligence.

00;27;52;19 - 00;27;54;10

It's the result of stupidity.

00;27;54;10 - 00;27;54;20

Right?

00;27;54;20 - 00;27;58;19

There is, you know, if you look at the

at the curve of intelligence,

00;27;58;19 - 00;27;59;27

if you want, right

00;27;59;27 - 00;28;04;10

there, is that point at which, you know,

the more you, the more intelligent

00;28;04;13 - 00;28;07;13

you become, the more positive

you have an impact on the world.

00;28;07;15 - 00;28;08;02

Right?

00;28;08;02 - 00;28;10;24

Until one certain point

where you're intelligent enough

00;28;10;24 - 00;28;13;24

to become a politician

or a corporate leader.

00;28;13;28 - 00;28;14;14

Okay.

00;28;14;14 - 00;28;18;17

And then but but you're not intelligent

enough to talk to your enemy,

00;28;18;20 - 00;28;19;00

right.

00;28;19;00 - 00;28;24;13

And when that happens,

that's when the impact dips to negative.

00;28;24;18 - 00;28;29;20

And that's the actual reason why

we are in so much pain in the world today.

00;28;29;23 - 00;28;30;04

Right.

00;28;30;04 - 00;28;33;24

But if you continue,

if you continue that curve,

00;28;33;27 - 00;28;38;20

intelligence, superior intelligence

by definition, is all juristic.

00;28;38;23 - 00;28;41;09

As a matter of fact,

this is in my writing.

00;28;41;09 - 00;28;45;15

I explain that as a, as a as a

as a property of physics if you want.

00;28;45;22 - 00;28;49;14

Because if you really understand

how the universe works,

00;28;49;15 - 00;28;53;18

you know, the everything we know

is the result of entropy, right?

00;28;53;22 - 00;28;55;21

The arrow of time

is the result of entropy.

00;28;55;21 - 00;28;57;02

The, you know, the

00;28;57;02 - 00;29;00;23

the current, universe in its current form

is the result of entropy.

00;29;00;23 - 00;29;03;23

Entropy is the tendency of the universe

to break down to,

00;29;03;29 - 00;29;07;06

to, to, to move from order to chaos

if you want.

00;29;07;12 - 00;29;10;01

That's the design of the universe, right?

00;29;10;01 - 00;29;13;01

The role of intelligence is that in

that universe

00;29;13;08 - 00;29;16;07

is to bring order

back to the chaos. Right.

00;29;16;10 - 00;29;21;12

And the most intelligent of all that

try to bring that order,

00;29;21;18 - 00;29;25;01

try to do it

in the most efficient way, right.

00;29;25;06 - 00;29;28;27

And the most efficient way

does not ever involve waste of waste

00;29;28;27 - 00;29;33;06

of resources, waste of lives,

you know, escalation of conflicts,

00;29;33;09 - 00;29;37;19

you know, consequences that lead

00;29;37;19 - 00;29;40;18

to further conflicts in the future

and so on and so forth.

00;29;40;24 - 00;29;44;24

And so in my mind,

when we completely hand over toy to AI,

00;29;44;25 - 00;29;50;14

which in my assessment is going to be 5

to 7 years, maybe 12 years at most, right?

00;29;50;20 - 00;29;53;16

There will be one general that will tell,

00;29;53;16 - 00;29;57;17

you know, it's his AI army

to go and kill a million people.

00;29;57;20 - 00;30;00;01

And the AI will go like,

why are you so stupid?

00;30;00;01 - 00;30;05;10

Like why I can talk to the other

AI in a microsecond and save everyone.

00;30;05;10 - 00;30;06;14

All of that.

00;30;06;14 - 00;30;09;14

You know, madness, right?

00;30;09;21 - 00;30;12;27

This is very anti-capitalist.

00;30;13;00 - 00;30;15;27

And so I sometimes when I warn about this,

00;30;15;27 - 00;30;20;27

I worry that the capitalists will hear me

and change the tactics right.

00;30;21;00 - 00;30;23;26

But but in reality,

00;30;23;26 - 00;30;25;18

it's it is inevitable.

00;30;25;18 - 00;30;29;16

Even if they do, it's inevitable

that, you know, we will hit

00;30;29;16 - 00;30;33;06

the second dilemma where everyone will

well have to go to AI.

00;30;33;09 - 00;30;34;21

Right? And it's inevitable.

00;30;34;21 - 00;30;36;09

I call it trusting intelligence.

00;30;36;09 - 00;30;41;01

That section of the book, it's inevitable

that, when we hand over to

00;30;41;06 - 00;30;45;19

to a superior intelligence,

it will not behave as stupidly as we do.

00;30;45;22 - 00;30;49;17

My prediction is that just like the first

Renaissance evolved, our understanding

00;30;49;17 - 00;30;53;16

what the social contract could look like

introduced enlightenment thought,

00;30;53;16 - 00;30;57;08

which led to, among other things,

democracy and new forms of government.

00;30;57;15 - 00;31;01;09

My prediction is that the next 50

to 100 years outside of novel sciences

00;31;01;09 - 00;31;05;12

will see the the biggest changes

in understanding of social contract.

00;31;05;15 - 00;31;09;16

And, you know, Sam has talked about this

a lot of people have talked about this.

00;31;09;19 - 00;31;13;22

In a world where work becomes

less critical to actually running society,

00;31;13;22 - 00;31;17;07

where value creation gets less expensive,

00;31;17;10 - 00;31;19;20

redefining how society should work

00;31;19;20 - 00;31;22;29

is going to require a bunch of people

to think about it.

00;31;22;29 - 00;31;24;23

And a bunch of like, quite honestly,

00;31;24;23 - 00;31;27;09

conflict within governments

to redefine those things.

00;31;27;09 - 00;31;32;05

And so if you're looking at 50

to 100 years out, my bold prediction is

00;31;32;08 - 00;31;37;04

new government like it truly like

democracy may not be the final state.

00;31;37;07 - 00;31;38;24

And we were probably destined

for something.

00;31;38;24 - 00;31;40;08

And by the way, I'm a free market.

00;31;40;08 - 00;31;43;08

Or capitalism might not even be

the free market solution.

00;31;43;10 - 00;31;46;27

We just don't know yet, because imagining

these things just will require

00;31;46;27 - 00;31;49;27

major updates

to how we understand the universe to work.

00;31;50;00 - 00;31;53;18

And overcoming that conflict is going

to is going to take a lot of work.

00;31;53;18 - 00;31;56;18

Now, the one other thing I'll say,

and we can come back to this, or

00;31;56;18 - 00;32;00;06

you can expand on this when people ask me,

what's the next great conflict?

00;32;00;06 - 00;32;03;05

I don't think it's between two nations,

I really don't.

00;32;03;05 - 00;32;05;01

I think we have reached

the sort of this flat

00;32;05;01 - 00;32;08;01

earth point where it's like, really

not in any nations interest,

00;32;08;06 - 00;32;11;06

especially nuclear equipped nations,

to fight.

00;32;11;11 - 00;32;14;05

And a hot war would just be, like,

so untenable.

00;32;14;05 - 00;32;17;09

You know, we would just

I don't think anyone wants it.

00;32;17;12 - 00;32;20;12

I do think that there is a future conflict

between people and the state.

00;32;20;18 - 00;32;24;05

I think there is a world where we wake up

in 20, 30 and 40 years and we go, oh,

00;32;24;05 - 00;32;27;05

we have all the things that the state

has been promising us.

00;32;27;06 - 00;32;29;06

It's just not the state that delivered it

right.

00;32;29;06 - 00;32;30;20

It's technology.

00;32;30;20 - 00;32;32;02

And that's going to be

one of these moments

00;32;32;02 - 00;32;36;01

where people go, I wonder why

I'm paying 50% taxes to a body

00;32;36;01 - 00;32;40;02

that, you know, doesn't

actually produce value anymore.

00;32;40;02 - 00;32;44;13

And so there's a whole other thing there,

which is like this introduces

00;32;44;13 - 00;32;46;00

this idea of a new form of government.

00;32;46;00 - 00;32;47;22

I think we get there

00;32;47;22 - 00;32;50;06

because a lot of people

are going to be like, wait a second, what?

00;32;50;06 - 00;32;53;25

Why are we being governed

in such a way that like,

00;32;53;25 - 00;32;57;06

it doesn't allow that,

you know, the technology to serve us?

00;32;57;10 - 00;33;00;12

One of my fears around AI, it's

00;33;00;12 - 00;33;03;19

nothing to do with laser eyed robots

or anything like that.

00;33;03;19 - 00;33;07;24

It's the consolidation in the hands

of the same few tech companies

00;33;07;24 - 00;33;11;11

that have been dominant,

for the last decade or two.

00;33;11;12 - 00;33;12;04

You know, it's funny.

00;33;12;04 - 00;33;16;22

So I started this book right

the end of 2022, started 2023.

00;33;16;25 - 00;33;19;25

And I went in search of the next Google,

the next meta.

00;33;19;28 - 00;33;22;16

And, you know,

I ended up concluding, like,

00;33;22;16 - 00;33;27;00

I fear that the next Google is Google in

AI and the next meta is meta.

00;33;27;03 - 00;33;28;24

You know that. Yeah, big.

00;33;28;24 - 00;33;30;14

This stuff is really expensive.

00;33;30;14 - 00;33;34;04

When I first started, you know,

people were talking about, you know,

00;33;34;06 - 00;33;39;17

millions, tens of millions

to train, fine tune and operate

00;33;39;17 - 00;33;43;27

these chatbots, large language models

where whatever you want to call it.

00;33;43;27 - 00;33;48;16

And the same with text to video,

audio to, a text audio.

00;33;48;19 - 00;33;53;05

By the time I was done

reporting at the end of 2024,

00;33;53;05 - 00;33;57;27

May 2024, it was hundreds of millions,

if not billions.

00;33;57;27 - 00;34;02;18

And Dario Modi for from anthropic

they do clogged the chat bot clogged.

00;34;02;21 - 00;34;05;11

You know he's

estimating that they're going to need $100

00;34;05;11 - 00;34;08;28

billion by 2027 to train these things.

00;34;09;01 - 00;34;11;22

And so who has that kind of money?

00;34;11;22 - 00;34;13;18

You know, Google, Microsoft.

00;34;13;18 - 00;34;16;12

They have 100 billion

or so laying around in cash.

00;34;16;12 - 00;34;21;20

But if you have to raise $100 billion,

or even if it's only, you know,

00;34;21;23 - 00;34;25;00

three, five, $10 billion, well,

a large venture capital outfit

00;34;25;00 - 00;34;29;14

in Silicon Valley has $1 billion,

all told, in a fund.

00;34;29;17 - 00;34;30;24

And so we're talking about billions.

00;34;30;24 - 00;34;35;04

And so that's one way

this is, way to to big tech.

00;34;35;04 - 00;34;36;19

And the other is data. Right.

00;34;36;19 - 00;34;40;13

This is really central to the,

the remedies that government is now

00;34;40;14 - 00;34;43;24

talking to remedies

for the Google antitrust trial.

00;34;44;00 - 00;34;49;15

You know, a quarter federal judge found

that, you know, Google is a monopolist.

00;34;49;16 - 00;34;52;03

It abused its power.

So now what should we do?

00;34;52;03 - 00;34;57;03

And a lot of the discussion,

I think rightfully is around the data.

00;34;57;06 - 00;35;01;16

You know, OpenAI approached

Google and said, hey, can we lease

00;35;01;16 - 00;35;05;24

we can we kind of buy access to your data

and they said, no.

00;35;05;27 - 00;35;08;04

And that's a huge advantage.

00;35;08;04 - 00;35;10;09

who are the people

who and or organizations

00;35;10;09 - 00;35;14;07

I guess that need to be on their toes

in this kind of changing world?

00;35;14;10 - 00;35;16;18

I think studios should be very worried.

00;35;16;18 - 00;35;21;11

I think anyone who's been an intermediary

for a long period of time, yeah.

00;35;21;15 - 00;35;24;22

Who's sort of been responsible

for the financing

00;35;24;22 - 00;35;28;02

or like the middleman deals

should be very concerned.

00;35;28;02 - 00;35;29;21

And that's not just the case in Hollywood.

00;35;29;21 - 00;35;30;28

I think that's the case across

00;35;30;28 - 00;35;36;08

every industry is being disintermediated

by these technologies.

00;35;36;11 - 00;35;37;03

And it's making

00;35;37;03 - 00;35;40;03

everything cheaper

and more easily accessible.

00;35;40;03 - 00;35;42;17

So I don't know what the studios

are going to do.

00;35;42;17 - 00;35;42;27

I hope

00;35;42;27 - 00;35;46;25

that they become really good at curating,

because we are going to have a problem

00;35;46;25 - 00;35;50;23

with noise as a result

of making everything cheaper and faster,

00;35;50;23 - 00;35;52;28

and will all of a sudden

everyone is going to be making content.

00;35;52;28 - 00;35;55;03

A lot of that's going to be very bad.

00;35;55;03 - 00;35;58;17

So who is it that has sort of the taste

making abilities

00;35;58;17 - 00;35;59;28

to curate the best content

00;35;59;28 - 00;36;04;01

and deliver it in a way that is really

appealing to large audiences.

00;36;04;01 - 00;36;07;01

So maybe that'll be the streamers,

maybe that'll be the studios,

00;36;07;01 - 00;36;09;28

but someone's going to have to win

at least that game.

00;36;09;28 - 00;36;11;01

I'm thinking about YouTube. Right.

00;36;11;01 - 00;36;15;12

Because YouTube is like, you know, this

this sea changing platform are the tech

00;36;15;12 - 00;36;20;01

companies becoming too powerful

here, like, is there I don't know.

00;36;20;01 - 00;36;21;18

Is there a dystopian risk.

00;36;21;18 - 00;36;24;20

What do you see the the changing role of,

you know, the technology companies

00;36;24;20 - 00;36;27;19

who like own the platforms as here?

00;36;27;25 - 00;36;31;11

Yeah, it's such a complicated question

and my answer

00;36;31;11 - 00;36;32;20

will probably be a little vague.

00;36;32;20 - 00;36;36;16

I think it's both and I absolutely see

the dystopian version of all this.

00;36;36;16 - 00;36;38;05

We're already living in it, right?

00;36;38;05 - 00;36;41;24

I mean, we're all addicted

to our smartphones.

00;36;41;26 - 00;36;45;14

And these social media apps

that are designed

00;36;45;14 - 00;36;48;27

to keep our attention

for as long as possible.

00;36;49;00 - 00;36;54;02

The algorithm is creating a feedback loop

around the types of content people want,

00;36;54;02 - 00;36;57;24

and that is also informing

what content creators are making.

00;36;57;27 - 00;37;02;20

And in many ways, you're seeing this,

I think sort of race to the bottom,

00;37;02;23 - 00;37;04;28

both in content and in storytelling.

00;37;04;28 - 00;37;06;16

Course, there's good stuff out there.

00;37;06;16 - 00;37;08;12

I don't want to say everything is bad.

00;37;08;12 - 00;37;10;08

There are plenty of really inspiring

creators

00;37;10;08 - 00;37;14;03

doing amazing things,

but there's also just,

00;37;14;06 - 00;37;18;07

you know, there's there are now hundreds

of thousands of creators

00;37;18;07 - 00;37;21;20

dedicated to teaching other people

how to grab attention.

00;37;21;23 - 00;37;24;22

You know, how to get someone

to click on your video

00;37;24;27 - 00;37;28;20

and and stay,

stay on you for longer than three seconds.

00;37;28;20 - 00;37;31;26

And they're right there

boiling this down into a science.

00;37;31;29 - 00;37;36;27

I think in the short term,

for as long as the age of,

00;37;37;00 - 00;37;40;12

of augmented intelligence is upon

us, those who cooperate

00;37;40;12 - 00;37;44;01

fully with AI and master

it are going to be winners.

00;37;44;04 - 00;37;46;00

There's absolutely no doubt about that.

00;37;46;00 - 00;37;48;22

Right.

00;37;48;25 - 00;37;53;29

Also, those who.

00;37;54;02 - 00;37;56;14

Excel

00;37;56;14 - 00;37;59;27

in the rare skill of human connection

00;38;00;02 - 00;38;02;21

will be winners, right?

00;38;02;21 - 00;38;07;04

Because I can sort of almost foresee

00;38;07;04 - 00;38;11;20

an immediate knee jerk reaction

to let's hand over everything to AI.

00;38;11;23 - 00;38;12;10

Right?

00;38;12;10 - 00;38;16;00

You know, I, I think the greatest example

is call centers, where, you know,

00;38;16;02 - 00;38;19;12

I get really frustrated

when I get an eye on a call center.

00;38;19;12 - 00;38;23;02

It's almost like your organization

is telling me they don't care enough.

00;38;23;05 - 00;38;23;25

Right?

00;38;23;25 - 00;38;27;08

And, and and, you know,

the idea here is I'm not underestimating

00;38;27;08 - 00;38;31;05

the value that an AI brings, but one,

they're not good enough yet.

00;38;31;08 - 00;38;31;22

Right.

00;38;31;22 - 00;38;37;14

And two, shouldn't I have I mean,

I wish you had realized that I can do

00;38;37;14 - 00;38;40;22

all of the mundane tasks

that made your call center agent

00;38;40;22 - 00;38;45;06

frustrated so that the call center agent

is actually nice to me, right?

00;38;45;12 - 00;38;49;24

So, so in the short term, I believe those

who there are three winners.

00;38;49;29 - 00;38;53;04

One is the

is the one that cooperate fully with AI.

00;38;53;08 - 00;38;56;16

The second is the one that, you know,

00;38;56;16 - 00;39;00;04

basically understands, human skills.

00;39;00;07 - 00;39;00;16

Right.

00;39;00;16 - 00;39;05;25

And human connection, on every front,

by the way, as, as AI replaces love and,

00;39;05;25 - 00;39;08;17

you know, tries to approach loneliness

and so on,

00;39;08;17 - 00;39;12;13

the ones that will actually go out

and meet girls who are going to be nicer.

00;39;12;16 - 00;39;13;01

Right?

00;39;13;01 - 00;39;16;06

They're going to be more attractive

if you want.

00;39;16;09 - 00;39;20;25

And then finally, I think the ones

that can parse out the truth.

00;39;20;28 - 00;39;21;12

Right.

00;39;21;12 - 00;39;24;07

So, so

what is one of the one of the sections

00;39;24;07 - 00;39;27;15

I wrote, so far,

published so far in my life is,

00;39;27;18 - 00;39;31;03

is a section that I called

The Age of Mind Manipulation.

00;39;31;06 - 00;39;34;13

And you'll be surprised that,

00;39;34;13 - 00;39;39;02

perhaps the skill,

that AI has acquired most,

00;39;39;05 - 00;39;42;23

in the in its early

years was to manipulate human minds.

00;39;42;26 - 00;39;45;11

Through social media.

00;39;45;11 - 00;39;47;25

And so and so my feeling is that,

00;39;47;25 - 00;39;53;15

there is a lot that you see today

that is not true.

00;39;53;18 - 00;39;57;00

Okay, that's not just fake videos,

which is, you know, the,

00;39;57;03 - 00;39;59;26

flamboyant example of,

00;39;59;26 - 00;40;03;10

of, of deepfake the,

00;40;03;10 - 00;40;06;12

the there is a lot that you see today

that is not true.

00;40;06;12 - 00;40;12;14

That comes into things like,

the bias of your feet.

00;40;12;17 - 00;40;13;02

Right?

00;40;13;02 - 00;40;16;26

If you're if you're from one side

or another of a conflict,

00;40;16;29 - 00;40;20;20

the, the AI of the internet would make

you think

00;40;20;20 - 00;40;24;14

that your view is the only right view

that everyone agrees, right?

00;40;24;18 - 00;40;27;03

You know, if you're a flat earther,

00;40;27;06 - 00;40;27;24

everyone.

00;40;27;24 - 00;40;28;28

It's like if someone tells you.

00;40;28;28 - 00;40;31;16

But is there any possibility

it's not flat?

00;40;31;16 - 00;40;35;06

You'll say, come on, everyone on

the internet is talking about it, right?

00;40;35;09 - 00;40;39;20

And, and I and I think the, the, the very,

very, very eye opening difference

00;40;39;20 - 00;40;43;20

which most people don't recognize

is, you know, I've had the privilege

00;40;43;20 - 00;40;48;28

of starting half of Google's businesses

worldwide and, and, you know,

00;40;49;01 - 00;40;54;18

got the internet and e-commerce and Google

to around 4 billion people.

00;40;54;21 - 00;40;58;02

And in Google, that wasn't a question

of opening a sales office.

00;40;58;02 - 00;41;01;28

That was really a deep

question of engineering, where you build

00;41;01;28 - 00;41;05;29

a product that understands the internet,

that improves the quality of the internet,

00;41;06;02 - 00;41;11;15

to the point where Bangladeshis

have access to democracy of information.

00;41;11;15 - 00;41;14;22

That's a massive contribution, right?

00;41;14;25 - 00;41;16;02

The thing is,

00;41;16;02 - 00;41;20;28

if you had asked Google

at any point in time until today,

00;41;21;01 - 00;41;24;00

any question, Google

would have responded to you

00;41;24;00 - 00;41;27;22

with a million possible

answers in terms of links

00;41;27;25 - 00;41;31;14

and said, go make up your mind

what you think is true, right?

00;41;31;17 - 00;41;35;00

If you ask ChatGPT today,

it gives you one answer

00;41;35;03 - 00;41;38;21

right and positions

it as the ultimate truth, right?

00;41;38;24 - 00;41;42;28

And it's so risky

that we humans accept that

00;41;43;02 - 00;41;46;06

do we really need

to, you know, keep pushing this forward

00;41;46;06 - 00;41;49;26

or do we have more than enough technology

here to, to keep us busy for the next

00;41;49;26 - 00;41;50;25

5 or 10 years?

00;41;50;25 - 00;41;52;29

How how do those two interplay.

00;41;52;29 - 00;41;54;24

More than enough technology.

00;41;54;24 - 00;41;57;05

That's I'm

I'm so glad you brought that up.

00;41;57;05 - 00;41;59;15

It's so funny

because I mean, I get it right.

00;41;59;15 - 00;42;03;17

I get the idea that, you know,

00;42;03;20 - 00;42;07;07

I get the idea that every company sort of

like the, the Google's, the Microsoft,

00;42;07;07 - 00;42;11;01

the OpenAI's, the anthropic, the,

you know, Meta's AI, etc.

00;42;11;04 - 00;42;13;18

they all want to be kind of atop

the leaderboard. Right.

00;42;13;18 - 00;42;15;05

And that for the best tech.

00;42;15;05 - 00;42;16;27

And believe me, I totally get it.

00;42;16;27 - 00;42;19;28

And it's, you know, but I actually,

you know, if you think about it,

00;42;19;29 - 00;42;22;10

if it's one of these things

where if you watch a Jeep commercial

00;42;22;10 - 00;42;23;17

or a Range Rover commercial,

00;42;23;17 - 00;42;25;28

like what are those Jeeps and Range Rovers

doing right?

00;42;25;28 - 00;42;28;11

They are doing things

like they are going over mountains.

00;42;28;11 - 00;42;31;06

These are people

that are like stuck in flowing rivers

00;42;31;06 - 00;42;33;01

and there's like a hippo

coming after them.

00;42;33;01 - 00;42;35;15

And they have six people in the car

and all, they're like,

00;42;35;15 - 00;42;38;18

oh my God, you've got us over there

doing unbelievable things.

00;42;38;21 - 00;42;41;00

Meanwhile,

I live in New Canaan, Connecticut.

00;42;41;00 - 00;42;43;23

There's Range Rovers and jeeps

all over the place. What are they driving?

00;42;43;23 - 00;42;46;17

They're driving over paved roads. Right?

00;42;46;17 - 00;42;47;04

They are.

00;42;47;04 - 00;42;51;10

They're driving from their home, their

three acre Hummer to to the train station.

00;42;51;13 - 00;42;54;04

So why are we all buying these things?

Right.

00;42;54;04 - 00;42;56;26

Like I have an Apple Watch on right now

that can go like 100m,

00;42;56;26 - 00;42;59;21

underwater in 18,000ft.

Do you think I'm ever doing that?

00;42;59;21 - 00;43;01;25

No, it's a it's sort of a feel like a.

00;43;01;25 - 00;43;05;00

Oh, but it could,

which means it's the best.

00;43;05;00 - 00;43;05;12

Right?

00;43;05;12 - 00;43;08;24

Whereas what I would say is that if

we just put everybody and we knew

00;43;08;24 - 00;43;14;11

everybody was going to use it at 3.5,

which was, you know, almost the original,

00;43;14;11 - 00;43;16;27

or close to the original model

that came out two years ago.

00;43;16;27 - 00;43;18;17

And if everybody actually using it,

00;43;18;17 - 00;43;21;10

you know, 20 times

a day, we'd be much further on.

00;43;21;10 - 00;43;24;15

So I get the tech, I appreciate the tech,

and I'm all over the tech.

00;43;24;15 - 00;43;27;15

I'm posting about it all the time

on LinkedIn, everything like that.

00;43;27;15 - 00;43;30;01

But to your point, your exact point,

the reality is

00;43;30;01 - 00;43;31;19

we have not caught up with the tech.

00;43;31;19 - 00;43;33;23

to be a winner in this new world,

00;43;33;23 - 00;43;37;19

you really have to learn to parse out

what is true and what is fake.

00;43;37;22 - 00;43;41;14

You really have to have the ability

to parse out what the media is telling you

00;43;41;18 - 00;43;44;19

to serve their own agendas

and what they're telling you.

00;43;44;19 - 00;43;46;09

That is actually true.

00;43;46;09 - 00;43;51;04

You know, you have to parse out

what actually happened versus opinion,

00;43;51;07 - 00;43;56;01

you know, what actually is

the truth versus the shiny headline.

00;43;56;04 - 00;43;59;04

And, and this is now going to be much

00;43;59;08 - 00;44;03;08

more potent

with artificial intelligence in charge,

00;44;03;14 - 00;44;06;11

because they have mastered

human manipulation.

00;44;06;11 - 00;44;08;03

AI keeps getting more and more intense.

00;44;08;03 - 00;44;11;12

Actually, as one would expect,

00;44;11;15 - 00;44;13;29

approaching this singularity

and all that, right?

00;44;13;29 - 00;44;18;01

So, I mean, it is it is interesting

to see it all happening finally.

00;44;18;04 - 00;44;20;29

I think progress is quite

00;44;21;02 - 00;44;21;21

amazing.

00;44;21;21 - 00;44;25;15

And looks exactly like you would think

00;44;25;15 - 00;44;30;09

if for in the last few years before

a breakthrough, the AGI and singularity.

00;44;30;12 - 00;44;33;25

So so it sounds like you're still pretty

bullish that we're you know marching I'm.

00;44;33;25 - 00;44;34;25

Super bullish man.

00;44;34;25 - 00;44;39;00

You know before literally before breakfast

this morning

00;44;39;03 - 00;44;41;21

I made like

00;44;41;21 - 00;44;45;17

ten Python programs

to test versions of some AI algorithm

00;44;45;17 - 00;44;49;09

I made up just by vibe

coding and Llvm platforms.

00;44;49;14 - 00;44;53;12

But before we had these tools,

each of those would have taken me

00;44;53;13 - 00;44;54;28

half a day, right?

00;44;54;28 - 00;44;57;05

So I mean, it sped up

00;44;57;05 - 00;44;59;02

prototyping research ideas

00;44;59;02 - 00;45;03;04

by a factor of 20 to 50 or something,

right?

00;45;03;05 - 00;45;09;29

I mean, and that that's tools that we have

now that are not remotely AGI,

00;45;09;29 - 00;45;13;10

they're just very useful, useful research

assistants.

00;45;13;10 - 00;45;17;10

But but we are at the

point where the AI tooling

00;45;17;13 - 00;45;19;20

is helping us

00;45;19;20 - 00;45;21;10

develop AI faster. Right.

00;45;21;10 - 00;45;25;06

And that is exactly what you would think

in the endgame period

00;45;25;06 - 00;45;27;02

before a singularity.

00;45;27;02 - 00;45;29;14

Well, and that can create a snowball

effect, right?

00;45;29;14 - 00;45;31;12

If it's helping us research

00;45;31;12 - 00;45;34;22

itself faster or any of these spaces

faster than, you know, that's.

00;45;34;22 - 00;45;36;13

Doing, it's doing that right now.

00;45;36;13 - 00;45;38;09

Yeah. I mean, that is that is

00;45;38;09 - 00;45;42;06

that is why we were able

to see the pace that we now see.

00;45;42;09 - 00;45;43;04

Yeah.

00;45;43;04 - 00;45;45;29

So, so, you know,

maybe just to take a step back then,

00;45;45;29 - 00;45;49;18

I mean artificial general intelligence,

this is a phrase that, you know,

00;45;49;18 - 00;45;54;01

you coined over a decade ago and has been

getting a lot of press lately,

00;45;54;04 - 00;45;56;01

in addition to superintelligence.

00;45;56;01 - 00;46;00;18

And so I wanted to ask you maybe

just to do a little bit of table setting,

00;46;00;21 - 00;46;01;04

how do you

00;46;01;04 - 00;46;04;04

define artificial general intelligence?

00;46;04;06 - 00;46;06;07

And, you know, why is why is it important?

00;46;06;07 - 00;46;07;17

Why does it matter?

00;46;07;17 - 00;46;09;27

And how does it differ, if at all?

00;46;09;27 - 00;46;13;28

You know, practically from something

like superintelligence.

00;46;14;01 - 00;46;17;08

So informally,

00;46;17;11 - 00;46;20;07

what we mean by AGI tends to be

00;46;20;07 - 00;46;24;12

the ability to generalize

00;46;24;15 - 00;46;26;13

roughly as well as people can.

00;46;26;13 - 00;46;29;13

And so to make leaps

beyond what you've been taught

00;46;29;13 - 00;46;32;26

and what you've been programed for,

to make those leaps,

00;46;32;29 - 00;46;34;17

you know roughly as well as people.

00;46;34;17 - 00;46;38;09

And that's that's an informal concept.

00;46;38;09 - 00;46;41;02

I mean, I mean, it's

not it's not a mathematical concept.

00;46;41;02 - 00;46;45;15

There's there's a mathematical theory

00;46;45;18 - 00;46;47;06

of general intelligence.

00;46;47;06 - 00;46;51;28

And it more deals with like,

what does it mean to be really,

00;46;51;28 - 00;46;53;14

really, really, really intelligent.

00;46;53;14 - 00;46;59;01

Like it's you can look at general

intelligence as the ability to achieve

00;46;59;04 - 00;47;03;29

arbitrary computable goals

and arbitrary computable environments. And

00;47;04;02 - 00;47;05;26

if you look at the abstract

00;47;05;26 - 00;47;11;20

math definition of general intelligence,

you conclude humans are not very far

00;47;11;20 - 00;47;15;16

along like, I cannot even run the maze in

00;47;15;16 - 00;47;19;04

750 dimensions, you know,

00;47;19;07 - 00;47;22;07

let alone prove a randomly generated

00;47;22;09 - 00;47;25;06

math theorem of length 10,000 characters.

00;47;25;06 - 00;47;28;07

I mean, I mean, we're we are adapted to do

00;47;28;07 - 00;47;31;14

the things that we evolved to do in our

00;47;31;17 - 00;47;32;12

environment, right?

00;47;32;12 - 00;47;35;21

We're not we're not utterly general

00;47;35;24 - 00;47;36;12

systems.

00;47;36;12 - 00;47;39;25

So, I mean, superintelligence

00;47;39;28 - 00;47;45;11

is also a very informally defined concept,

but it basically means is a system

00;47;45;11 - 00;47;50;09

whose general intelligence is way above

the human level of general intelligence.

00;47;50;09 - 00;47;55;14

So it can it can make creative leaps

beyond what it knows

00;47;55;17 - 00;47;58;05

way, way better

than, than the person can write.

00;47;58;05 - 00;48;02;27

And, I mean, it's pretty clear

00;48;03;00 - 00;48;04;09

that's possible.

00;48;04;09 - 00;48;08;16

I mean, just as we're not the fastest

running at highest jumping possible

00;48;08;16 - 00;48;13;26

creatures, we're probably not the smartest

thinking possible creatures.

00;48;13;29 - 00;48;19;07

And we can see examples of human

stupidity, like, around us every day

00;48;19;07 - 00;48;22;27

or even very smart people like I can hold.

00;48;23;00 - 00;48;26;13

I'm pretty clever, but I can no 1015

00;48;26;13 - 00;48;29;13

things in my memory

at one time without getting confused.

00;48;29;13 - 00;48;33;11

Now some autistic people can do better,

but I mean,

00;48;33;14 - 00;48;37;13

you know,

there are many limitations of being

00;48;37;16 - 00;48;41;07

a human brain, and it seems clear

00;48;41;10 - 00;48;44;05

some physical system

could do better than that.

00;48;44;05 - 00;48;47;11

And then the

the relation between human level AGI

00;48;47;11 - 00;48;50;17

and RSI is interesting

because it seems like

00;48;50;20 - 00;48;54;29

once you get a human level AGI,

like a computer system,

00;48;55;02 - 00;48;58;19

that on the one hand can generalize

and imagine and create

00;48;58;19 - 00;49;02;23

as well as the person on the other hand

is inside a computer.

00;49;02;26 - 00;49;07;14

It seems like that human level

AGI should pretty rapidly create

00;49;07;14 - 00;49;12;18

or become an aside, because, I mean,

it can look at its entire ram state.

00;49;12;18 - 00;49;14;07

It knows, oh, its source code,

00;49;14;07 - 00;49;18;00

it can copy itself and tweak itself

and run that copy on different machines

00;49;18;00 - 00;49;19;05

experimentally. Right.

00;49;19;05 - 00;49;25;15

So I mean, it seems like a human

level AGI will have much

00;49;25;18 - 00;49;26;24

greater ability to

00;49;26;24 - 00;49;30;23

self understand and self

modify the human level human,

00;49;30;23 - 00;49;35;00

which should should lead

should lead to ASI

00;49;35;03 - 00;49;37;23

fairly rapidly.

00;49;37;23 - 00;49;42;01

And you know, we've seen

in the commercial world some attempts

00;49;42;01 - 00;49;46;22

by business and marketing people

to fudge around with what is AGI.

00;49;46;25 - 00;49;49;17

But I mean,

I think within the research world,

00;49;49;17 - 00;49;52;01

the notion that an AGI

00;49;52;01 - 00;49;55;05

should be able to generalize

00;49;55;08 - 00;49;58;25

very well beyond its training data,

at least as well as people.

00;49;58;28 - 00;50;00;15

I think that's well recognized.

00;50;00;15 - 00;50;03;21

I mean, I've seen Sam

Altman has come out saying, well,

00;50;03;21 - 00;50;07;21

maybe something to do 95% of human jobs,

we should call it an AGI.

00;50;07;21 - 00;50;10;15

And I mean, you can

you can call it what you want.

00;50;10;15 - 00;50;11;05

It's fine.

00;50;11;05 - 00;50;16;06

But it is a different concept than having

human like generalization ability.

00;50;16;06 - 00;50;16;13

Right?

00;50;16;13 - 00;50;20;29

Like if you can do 95% of human jobs

by being trained in all of them,

00;50;20;29 - 00;50;25;22

I mean that that may be super,

super economically useful,

00;50;25;25 - 00;50;29;03

but it's different than being able to take

big leaps beyond your training that.

00;50;29;06 - 00;50;32;06

the technology's moving faster

than any other sector, faster

00;50;32;06 - 00;50;35;17

in the economy, fashion, the society

is moving faster and education's moving.

00;50;35;24 - 00;50;40;11

And if we truly want to understand

where humans play in that picture,

00;50;40;14 - 00;50;43;09

the fact that we're investing

everything we have in technology

00;50;43;09 - 00;50;47;16

has already indicated our preference

for technology over humans.

00;50;47;19 - 00;50;50;15

So that math has to balance out a bit.

00;50;50;15 - 00;50;53;26

We have to figure out

how do we invest so much more

00;50;54;01 - 00;50;57;01

into education, not so much less.

00;50;57;02 - 00;51;00;02

And until we do that,

we are going to be behind the ball.

00;51;00;02 - 00;51;02;03

We are going to have a target on our back

in many ways,

00;51;02;03 - 00;51;05;24

because if the paradigms don't change,

the technology just gets better.

00;51;05;27 - 00;51;07;29

We're going to suffer the consequences.

00;51;07;29 - 00;51;11;19

But if we put ourselves

front and center of that equation,

00;51;11;22 - 00;51;15;11

we have the chance and the opportunity

to figure that out.

00;51;15;14 - 00;51;17;08

Right? It's wow.

00;51;17;08 - 00;51;22;09

Yeah, it's it's as you said it like,

this is not

00;51;22;12 - 00;51;23;13

an incremental shift.

00;51;23;13 - 00;51;27;07

This is like a complete disruption

of the model from end to end.

00;51;27;10 - 00;51;28;09

Without a doubt.

00;51;28;09 - 00;51;31;22

And even for people who live and breathe

it like it's overwhelming for me,

00;51;31;25 - 00;51;34;08

I do this 24 seven,

I love it. I'm passionate about it.

00;51;34;08 - 00;51;36;00

I'm excited about where we're going

00;51;36;00 - 00;51;39;17

and net net,

I'm optimistic about the long term future,

00;51;39;20 - 00;51;43;09

but we are all pioneers right now,

whether we want to be or not.

00;51;43;12 - 00;51;46;26

And when people we've kind of bastardized

the term pioneer,

00;51;46;26 - 00;51;49;01

we've made it seem like,

oh, it's Richard Branson on the cover

00;51;49;01 - 00;51;52;01

of entrepreneur magazine with his billions

of dollars of success, like

00;51;52;02 - 00;51;54;21

he was a pioneer at one point in time.

00;51;54;21 - 00;51;58;00

But yeah, high nears

through really hard shit

00;51;58;00 - 00;52;01;14

and they go to places

where there's no infrastructure.

00;52;01;19 - 00;52;05;11

They suffer the consequences of,

you know, decisions that they didn't know

00;52;05;11 - 00;52;06;03

they'd have to make.

00;52;06;03 - 00;52;10;03

They are attacked by the environment

that they're in nature tries to kill them

00;52;10;03 - 00;52;11;14

in a number of different ways. Yeah.

00;52;11;14 - 00;52;15;06

And as a super resilient species,

we still make a way forward.

00;52;15;13 - 00;52;18;01

We construct the environment

after we figure it out.

00;52;18;01 - 00;52;19;16

You know, we might show up in Hawaii

00;52;19;16 - 00;52;23;12

with no shoes on and realize, oh, crap,

I'm not properly equipped for this.

00;52;23;15 - 00;52;26;06

And then we figure a way out that pattern.

00;52;26;06 - 00;52;30;06

Time to go from not knowing to

knowing can be really hard, painful,

00;52;30;06 - 00;52;35;19

and challenging, but the way we thrive

once we do is absolutely amazing.

00;52;35;22 - 00;52;39;10

So I would say that we are going

to have amazing things happen,

00;52;39;13 - 00;52;43;03

but we're also going to have to encounter

some really tough growing pains

00;52;43;06 - 00;52;45;08

individually

and collectively to get there.

00;52;45;08 - 00;52;48;08

So if anyone is saying otherwise, it's

absolutely smoke and mirrors

00;52;48;08 - 00;52;52;22

whether there's a threat

to creating this intelligence that,

00;52;52;28 - 00;52;55;28

you know, looks at this,

you know, kind of human pandemonium

00;52;56;00 - 00;52;57;02

and says, you know what?

00;52;57;02 - 00;53;01;02

You know, AI is taking the wheel

now, humans can't be trusted

00;53;01;06 - 00;53;02;16

with human affairs.

00;53;02;16 - 00;53;05;07

And this word that,

that we were so anchored on of choice.

00;53;05;07 - 00;53;06;03

And there's going to be.

00;53;06;03 - 00;53;08;06

It's almost inevitable.

00;53;08;06 - 00;53;10;03

And the AGI will be right.

00;53;10;03 - 00;53;13;13

I mean, and then

00;53;13;16 - 00;53;14;24

human

00;53;14;24 - 00;53;18;07

governance systems

become more like the student council

00;53;18;07 - 00;53;21;20

in my high school or something,

where, I mean,

00;53;21;23 - 00;53;24;23

I mean, I think,

00;53;24;23 - 00;53;28;01

if you set aside AGI,

00;53;28;04 - 00;53;31;02

I mean, we can develop better and better

bio weapons.

00;53;31;02 - 00;53;33;02

There will be no nano weapons.

00;53;33;02 - 00;53;35;16

I mean, cybersecurity

00;53;35;16 - 00;53;37;00

barely works, right?

00;53;37;00 - 00;53;42;13

So, I mean, I think I think,

00;53;42;16 - 00;53;44;07

It seems almost inevitable

00;53;44;07 - 00;53;47;07

that rational humans

00;53;47;12 - 00;53;49;18

would democratically choose

00;53;49;18 - 00;53;52;19

to put a compassionate AGI

00;53;52;26 - 00;53;56;23

in some sort of a governance role,

00;53;56;26 - 00;54;00;05

given what the alternatives

00;54;00;08 - 00;54;01;17

appear to be.

00;54;01;17 - 00;54;06;11

But the the kind of

00;54;06;14 - 00;54;07;26

goofball analogy

00;54;07;26 - 00;54;11;23

I've often given is the

the squirrels in Yellowstone Park.

00;54;11;23 - 00;54;14;05

Like we're sort of in charge of them.

00;54;14;05 - 00;54;17;03

We're not actually micromanaging

their lives.

00;54;17;03 - 00;54;17;13

Right?

00;54;17;13 - 00;54;20;15

Like we're we're not telling the squirrels

who to mate with or what

00;54;20;15 - 00;54;24;06

tree to tree to climb up or something

like that, right?

00;54;24;06 - 00;54;28;15

Where, you know, if there was

a massive war between the white tails

00;54;28;15 - 00;54;32;16

and the brown tailed squirrels

and there's massive squirrel slaughter,

00;54;32;19 - 00;54;37;08

we might somehow intervene and move some

of them across the river or something.

00;54;37;11 - 00;54;42;01

If there's a plague, we would go in

and given the medicine, the by and large,

00;54;42;04 - 00;54;44;29

we know that for them to be squirrels,

00;54;44;29 - 00;54;49;04

they need to regulate their own lives

in their in their squirrely way.

00;54;49;04 - 00;54;49;11

Right.

00;54;49;11 - 00;54;52;11

And so that that is what you would hope

00;54;52;11 - 00;54;56;02

from a beneficial superintelligence

like it would know that

00;54;56;05 - 00;54;59;14

people would feel disempowered

00;54;59;17 - 00;55;04;02

and unsatisfied, have their lives

00;55;04;05 - 00;55;05;05

and their governments

00;55;05;05 - 00;55;09;16

micromanaged by some, by some AI system.

00;55;09;16 - 00;55;13;14

So what what you would hope is

a beneficial AGI is kind of there

00;55;13;14 - 00;55;17;13

in the background as a safety mechanism.

00;55;17;16 - 00;55;20;27

If it would stop stupid words

from popping up all over the world

00;55;20;27 - 00;55;22;00

like we see right now.

00;55;22;00 - 00;55;25;29

I mean,

I think that would be quite beneficial.

00;55;26;02 - 00;55;29;12

I don't see why we humans

need the AGI to decide.

00;55;29;12 - 00;55;34;14

Like, you know, what rights,

what rights do do children have?

00;55;34;14 - 00;55;35;20

Like what?

00;55;35;20 - 00;55;36;16

You know what?

00;55;36;16 - 00;55;39;01

How is the public school system

regulated or something?

00;55;39;01 - 00;55;43;05

There's

lots of lots of aspects of human life

00;55;43;08 - 00;55;47;15

that are going to be better dealt with

by humans collectively making decisions

00;55;47;18 - 00;55;51;22

for other humans with whom

they entered into a social contract.

00;55;51;22 - 00;55;52;02

Right?

00;55;52;02 - 00;55;56;27

So, I mean, I, I think, anyway,

there are clearly beneficial

00;55;57;00 - 00;55;57;23

avenues.

00;55;57;23 - 00;56;03;16

I mean, there's also many dystopic avenues

which we've all heard

00;56;03;19 - 00;56;07;16

heard plenty about,

00;56;07;19 - 00;56;08;26

I don't see any reason

00;56;08;26 - 00;56;12;26

why dystopian avenues are highly probable,

00;56;12;29 - 00;56;15;27

but I'm really more worried about what

00;56;15;27 - 00;56;18;24

nasty people do with early stage.

00;56;18;24 - 00;56;19;28

AGI is right.

00;56;19;28 - 00;56;24;23

I mean, I think there's a lot of possible

AI minds that could be built.

00;56;24;26 - 00;56;28;28

There's a lot of possible goals,

motivational and esthetic systems

00;56;28;28 - 00;56;33;06

that Agis could have.

00;56;33;09 - 00;56;35;22

I don't think we need to worry

that much about, like

00;56;35;22 - 00;56;38;22

the AGI is built to be compassionate,

loving and nice.

00;56;38;28 - 00;56;40;12

It's something everyone

00;56;40;12 - 00;56;43;10

then suddenly reverses and starts

slaughtering everyone, right?

00;56;43;10 - 00;56;47;02

I mean, it could happen,

but there's totally no reason to think

00;56;47;02 - 00;56;50;02

that's likely.

00;56;50;02 - 00;56;51;24

On the other hand,

00;56;51;24 - 00;56;55;28

the idea that some powerful party

with a lot of money

00;56;55;28 - 00;56;58;28

could try to build the smartest AGI

in the world

00;56;59;02 - 00;57;02;01

to promote their own interests

above everybody else's

00;57;02;01 - 00;57;06;23

and make everyone else fall into line

according to their will of that.

00;57;06;26 - 00;57;10;26

That's a very immediate and palpable

threat.

00;57;10;26 - 00;57;12;10

Right.

00;57;12;10 - 00;57;15;21

So. And that that

00;57;15;24 - 00;57;19;27

even if that doesn't affect

the ultimate superintelligence you get,

00;57;20;00 - 00;57;24;16

it could make things very unpleasant

for like five, ten,

00;57;24;16 - 00;57;29;21

20 years along the, the, the way which,

which matters a lot to us.

00;57;29;24 - 00;57;31;06

we have to remember that

00;57;31;06 - 00;57;34;25

all of these technologies

we're discussing are in their infancy.

00;57;34;28 - 00;57;39;03

And they're historically,

when you look at the advent of New,

00;57;39;06 - 00;57;41;19

particularly media forms of media,

00;57;41;19 - 00;57;44;01

it takes years for society to figure out

00;57;44;01 - 00;57;47;01

what they're for.

00;57;47;04 - 00;57;48;20

The telephone

00;57;48;20 - 00;57;52;22

for the first 25 years, the telephones

life, the telephone industry

00;57;52;22 - 00;57;56;21

actively tried to discourage people

from using it to

00;57;56;24 - 00;57;59;18

to gossip, to catch up with friends.

00;57;59;18 - 00;58;00;19

They thought it was a business.

00;58;00;19 - 00;58;03;08

Tool beneath the function

of the technology. Yeah.

00;58;03;08 - 00;58;07;26

This should not be shouldn't squander

this thing on chatting with your mom.

00;58;07;26 - 00;58;11;28

You should it. It's a business

tool and they actually,

00;58;12;01 - 00;58;14;06

like I said, actively discourage

people from.

00;58;14;06 - 00;58;17;29

They didn't realize what it was

until the 20s, which is, you know,

00;58;18;01 - 00;58;22;14

when does Alexander Graham Bell

telephone 1873.

00;58;22;17 - 00;58;24;08

And it's it's it's

00;58;24;08 - 00;58;29;17

the end of the 1920s

before they wake up to what it is. So

00;58;29;20 - 00;58;32;22

and the telephone is a bigger deal

than Facebook and Twitter.

00;58;32;23 - 00;58;34;10

Yeah.

00;58;34;10 - 00;58;38;13

But it strikes me that Facebook and

Twitter are, are still in their infancy.

00;58;38;13 - 00;58;39;19

They're really young.

00;58;39;19 - 00;58;42;13

Do I even it's quite possible

that if we had

00;58;42;13 - 00;58;45;27

this conversation five years from now,

00;58;46;00 - 00;58;48;19

not neither of you,

both of us would have only a dim

00;58;48;19 - 00;58;50;14

memory of this thing called Facebook.

00;58;50;14 - 00;58;53;27

Yeah, or Twitter or,

I don't know, or the opposite.

00;58;54;02 - 00;58;56;05

Yeah,

that it completely dominates your life.

00;58;56;05 - 00;58;57;26

I just don't think

00;58;57;29 - 00;58;59;10

the only confidence I have is

00;58;59;10 - 00;59;03;13

that we will be using these technologies

in unanticipated ways.

00;59;03;16 - 00;59;05;19

Yeah, in the future, but I

00;59;05;19 - 00;59;09;12

but no one can predict

what those unanticipated ways are.

00;59;09;15 - 00;59;13;15

I think I have a lot of confidence that

00;59;13;18 - 00;59;16;25

whatever employment dislocation

00;59;16;28 - 00;59;21;29

is caused by AI will be.

00;59;22;02 - 00;59;24;23

Will be short and

00;59;24;23 - 00;59;27;17

not painless,

but less painful than we think.

00;59;27;17 - 00;59;28;10

I think the gloom.

00;59;28;10 - 00;59;31;26

And I don't buy the doom

or the gloom and doom thing on it.

00;59;31;29 - 00;59;35;04

Yeah. Just think like

00;59;35;07 - 00;59;37;09

we always say this

every time something comes along.

00;59;37;09 - 00;59;40;16

It never pans out to yeah,

everyone has nothing to do.

00;59;40;17 - 00;59;42;04

It's become kind of Malthusian, right?

00;59;42;04 - 00;59;44;07

Like this wave is going to. Yeah.

00;59;44;07 - 00;59;46;25

And I think people have more ingenuity

than that.

00;59;46;25 - 00;59;53;08

And also

I think that we're probably a lot

00;59;53;11 - 00;59;55;24

further off from

00;59;55;24 - 00;59;58;16

truly transformative AI than we realize.

00;59;58;16 - 01;00;02;18

I just, I'm

a I'm on the kind of my expectations are.

01;00;02;25 - 01;00;09;02

But I also I also

sometimes believe that a lot of the most.

01;00;09;05 - 01;00;10;24

You know, revolutionary

01;00;10;24 - 01;00;13;24

uses of AI are some of its simplest ones

01;00;13;25 - 01;00;17;28

and just need doesn't need to be this

01;00;18;01 - 01;00;19;25

incredibly mind blowing

01;00;19;25 - 01;00;23;10

technological accomplishment

to make a difference in our lives.

01;00;23;13 - 01;00;26;24

Simply holding and organizing information

01;00;26;27 - 01;00;33;01

and standing at the ready

to give good answers to problems is huge.

01;00;33;01 - 01;00;36;01

I mean, if that's all it did,

it would be transformative.

01;00;36;02 - 01;00;39;18

things are going to look so different

in the next couple of years

01;00;39;18 - 01;00;42;20

unless you are radical with your thinking,

you will not be ready

01;00;42;20 - 01;00;44;12

for the disruptions

that are going to come.

01;00;44;12 - 01;00;48;07

middle management is also getting hit

really hard What that makes space

01;00;48;07 - 01;00;51;06

for is for people

to step into actual roles of leadership.

01;00;51;07 - 01;00;54;20

can imagine a world where there's angst.

01;00;54;23 - 01;00;57;28

If we're not looking forward

and we're still letting yesterday's

01;00;57;28 - 01;01;01;22

mental models collide with tomorrow's

technologies, that is how we lose.

01;01;01;22 - 01;01;04;20

we are all pioneers right now,

whether we want to be or not.

01;01;04;20 - 01;01;08;09

I would encourage organizations

to be radical with their thinking

01;01;08;12 - 01;01;10;25

and practical with their approach.

01;01;10;25 - 01;01;13;23

So there's there are too many people.

01;01;13;23 - 01;01;17;18

So you kind of need to break burning

all the ground, start fresh.

01;01;17;21 - 01;01;20;19

There's no enterprise

that says we're profitable.

01;01;20;19 - 01;01;21;28

We're doing just fine.

01;01;21;28 - 01;01;25;08

We want to disrupt that. Nobody says that.

01;01;25;11 - 01;01;29;07

But what I do think is,

unless you are radical with your thinking,

01;01;29;07 - 01;01;32;14

you will not be ready for the disruptions

that are going to come.

01;01;32;17 - 01;01;36;08

So these technological transformations

that happen at GPT level,

01;01;36;08 - 01;01;39;23

so general purpose technology

start at the infrastructure level.

01;01;39;26 - 01;01;42;28

So we've seen disruption with technology

and the technology that we use.

01;01;42;29 - 01;01;44;21

So electricity did the same thing.

01;01;44;21 - 01;01;47;06

And OpenAI did the same thing with GPT.

01;01;47;06 - 01;01;49;01

So we know now we're all using it.

01;01;49;01 - 01;01;51;24

But over time those disruptions move up

01;01;51;24 - 01;01;55;17

a level from infrastructure

to application to industry.

01;01;55;20 - 01;01;57;28

So if you're not

01;01;58;01 - 01;02;01;26

okay I guess I guess it is explosive.

01;02;01;29 - 01;02;04;11

But if you're not thinking radically

01;02;04;11 - 01;02;07;13

about the transformation that can happen

at each one of those levels

01;02;07;13 - 01;02;10;28

and also the transformation

that can happen to your industry,

01;02;11;01 - 01;02;14;06

and you're just focused on the data,

what you have now,

01;02;14;09 - 01;02;16;29

you're missing one of the critical shifts

of transformation in the business.

01;02;16;29 - 01;02;20;29

And there's a, theme

that's becoming more popular

01;02;20;29 - 01;02;23;29

right now

is going moving from insight to foresight.

01;02;24;00 - 01;02;28;16

And when everything is changing around

you, insights valuable.

01;02;28;16 - 01;02;32;09

It's how you create structure around

a business that you can take to market.

01;02;32;12 - 01;02;35;12

Foresight

is about how you avoid getting disrupted.

01;02;35;19 - 01;02;38;24

If we're not looking forward

and we're still letting yesterday's

01;02;38;24 - 01;02;43;00

mental models collide with tomorrow's

technologies, that is how we lose.

01;02;43;03 - 01;02;47;03

But if we are radical

with the way we think, with the ability

01;02;47;03 - 01;02;48;26

to test different business models,

01;02;48;26 - 01;02;52;16

put things to market faster

when we might not previously

01;02;52;19 - 01;02;55;16

get that data and that feedback loop

as fast as possible,

01;02;55;16 - 01;02;59;05

we're going to learn more

about that unexplored terrain way faster.

01;02;59;08 - 01;03;04;12

So I wouldn't say go and disrupt your

your $1 billion, you know, revenue line,

01;03;04;15 - 01;03;07;13

but you absolutely

should be incubating things that will

01;03;07;13 - 01;03;09;03

because there are hundreds

01;03;09;03 - 01;03;12;03

and eventually thousands of other startups

that are doing exactly that.

01;03;12;09 - 01;03;15;00

And you have no defense against that

if you're not thinking in that way.

01;03;15;00 - 01;03;17;29

So think radically approach practically.

01;03;17;29 - 01;03;19;23

So that next step goes, okay.

01;03;19;23 - 01;03;21;19

So what do we do to implement this.

01;03;21;19 - 01;03;24;11

Is it Tiger teams. Is it small skunkworks.

01;03;24;11 - 01;03;25;19

All of those are viable.

01;03;25;19 - 01;03;29;05

I do believe that

having in its transformation,

01;03;29;05 - 01;03;32;21

you need to find people who are leaning in

and are self-selecting

01;03;32;21 - 01;03;35;21

as the people who are like, I'm

all about this, I want to do this.

01;03;35;24 - 01;03;39;14

Don't try and convince a bunch of people

who might not be invested

01;03;39;14 - 01;03;41;26

in the nest

to be the first ones through the door.

01;03;41;26 - 01;03;43;13

They will be unenthusiastic about it.

01;03;43;13 - 01;03;45;22

They don't have the willpower

to get through the challenges.

01;03;45;22 - 01;03;47;00

It's going to be hard,

01;03;47;00 - 01;03;50;00

and they're going to fail a million times

before they get it right.

01;03;50;00 - 01;03;52;28

If they're not already passionate

about this,

01;03;52;28 - 01;03;55;12

they're going to stop

at the first sign of trouble.

01;03;55;12 - 01;03;58;19

Those people can be followers

of the people who lead the way.

01;03;58;19 - 01;04;00;03

It's not that they're irrelevant.

01;04;00;03 - 01;04;01;15

You need to find the people who are like,

01;04;01;15 - 01;04;03;13

I want to be

the person who kicks the door down.

01;04;03;13 - 01;04;06;16

I want the first person in the room,

and those are the ones you want to build

01;04;06;16 - 01;04;10;16

your teams around to, to think about

these things and build different ideas

01;04;10;19 - 01;04;12;01

and find the tinkerers.

01;04;12;01 - 01;04;14;05

Find the people who may not be

01;04;14;05 - 01;04;17;20

the developers or the engineers

who are already tinkering with the stuff.

01;04;17;23 - 01;04;20;23

There are so many people who are using AI

and building their own agents

01;04;20;23 - 01;04;23;19

or creating,

you know, side businesses on the weekends

01;04;23;19 - 01;04;25;04

who could also be resources for this.

01;04;25;04 - 01;04;29;22

And that's the culture that will create

new opportunities, new business models.

01;04;29;25 - 01;04;30;24

And they're going to learn

01;04;30;24 - 01;04;34;27

what these new paradigms will look like

by doing the work in that space

01;04;35;00 - 01;04;37;26

that then can be diffused

across the organization.

01;04;37;26 - 01;04;39;26

And that's the second most important part.

01;04;39;26 - 01;04;43;03

Once you have the knowledge,

do you have the infrastructure

01;04;43;03 - 01;04;46;07

set up to diffuse

that knowledge as fast as possible

01;04;46;13 - 01;04;49;08

and as thoroughly as possible

across the organization?

01;04;49;08 - 01;04;50;16

Otherwise, it just is.

01;04;50;16 - 01;04;53;12

Compartmentalize it. Compartmentalize it.

It dies on the vine.

01;04;53;12 - 01;04;55;25

I think that not enough companies

appreciate

01;04;55;25 - 01;05;00;07

that innovation demands waste.

01;05;00;10 - 01;05;01;12

But if you are doing

01;05;01;12 - 01;05;04;18

something that you've done before,

you know exactly how it's going to go,

01;05;04;21 - 01;05;08;01

then of course you can have these KPIs

01;05;08;01 - 01;05;12;06

that you know you're going to hit for sure

because you've already done it.

01;05;12;09 - 01;05;16;08

Now you're trying a completely new

technology, the completely new use case.

01;05;16;11 - 01;05;19;04

You have no idea if it's going to work.

01;05;19;04 - 01;05;21;02

You have to be willing to accept

01;05;21;02 - 01;05;24;20

that that might be time and effort

01;05;24;23 - 01;05;29;08

thrown, you know, burned

at the altar of innovation, so to speak.

01;05;29;11 - 01;05;31;26

Right? That

that is just the nature of innovation.

01;05;31;26 - 01;05;35;19

And I've had companies

come and consult with me

01;05;35;22 - 01;05;39;08

who they really wanted to be innovators.

01;05;39;10 - 01;05;45;00

But when I ask them,

so what is your actual tolerance

01;05;45;03 - 01;05;49;02

for getting no results back

after you invest in innovation?

01;05;49;06 - 01;05;52;16

For how much bandwidth

do you give your people to do things

01;05;52;16 - 01;05;57;19

that are very specific work product

that you expect from them?

01;05;57;19 - 01;06;02;02

Do you give them time

and space to to chase an idea?

01;06;02;05 - 01;06;03;23

And quite often the answer is no,

01;06;03;23 - 01;06;06;10

no we don't.

We have no tolerance for innovation.

01;06;06;10 - 01;06;09;09

We have absolutely no slack

for our people.

01;06;09;09 - 01;06;12;29

And we need every project

to be predictable.

01;06;13;02 - 01;06;16;05

Okay, if you're dealing with that,

you're just not going to be an innovator

01;06;16;05 - 01;06;18;12

or you're going to be

an accidental innovator

01;06;18;12 - 01;06;22;02

because you somehow accidentally hired

somebody who's going to essentially work

01;06;22;03 - 01;06;24;06

two jobs, the one you gave them, and then,

01;06;24;06 - 01;06;27;07

you know, the other one,

they'll spend nights in the office

01;06;27;07 - 01;06;30;26

and maybe they'll come up with something,

but there won't be a lot of these folks.

01;06;30;26 - 01;06;35;09

And, yeah,

that's not a great lottery ticket. So,

01;06;35;12 - 01;06;39;04

if you don't have that tolerance for

01;06;39;07 - 01;06;44;03

her, why, when you're trying to innovate,

01;06;44;06 - 01;06;45;20

just you have to be a follower.

01;06;45;20 - 01;06;49;25

It just wait for everybody else to to,

share how it's done and follow them.

01;06;49;29 - 01;06;53;14

When you say narrow AI,

what does that mean to you?

01;06;53;14 - 01;06;54;16

And is there, you know,

01;06;54;16 - 01;06;58;24

a threshold where it gets too broad

and that creates the risk for us.

01;06;58;27 - 01;07;03;19

So typically it's a system designed for

a specific purpose that can do one thing.

01;07;03;19 - 01;07;07;16

While it can play chess well,

it can do protein folding.

01;07;07;16 - 01;07;08;20

Well.

01;07;08;20 - 01;07;11;13

It's getting fuzzy when it becomes

01;07;11;13 - 01;07;14;27

a large neural network

with lots of capabilities.

01;07;14;27 - 01;07;19;19

So I think sufficiently advanced narrow

AI times to shift over

01;07;19;19 - 01;07;24;14

its more general capabilities, or

it can be quickly repurposed to do that,

01;07;24;17 - 01;07;27;10

but it's still a much better path forward

01;07;27;10 - 01;07;31;04

than feeding it all the data in the world

and seeing what happens.

01;07;31;04 - 01;07;31;13

Yeah.

01;07;31;13 - 01;07;36;21

So if you restrict your training data

to a specific domain, just play chess.

01;07;36;21 - 01;07;41;17

It's very unlikely to also be able

to do synthetic biology right.

01;07;41;20 - 01;07;44;11

Well, and it feels like we're very much

01;07;44;11 - 01;07;47;29

on the course of chess

and synthetic biology at the same time.

01;07;47;29 - 01;07;48;05

Right.

01;07;48;05 - 01;07;50;04

Is that is that your your kind of outlook

01;07;50;04 - 01;07;53;29

for where all the money is going

and what people are racing toward?

01;07;54;02 - 01;07;57;23

They explicitly saying that

superintelligence now they skipped AGI.

01;07;57;23 - 01;08;00;13

It's no longer even fun to talk about.

They directly.

01;08;00;13 - 01;08;04;13

When we have a superintelligence team,

we have superintelligence safety team.

01;08;04;16 - 01;08;06;01

You couldn't do it for any guy.

01;08;06;01 - 01;08;08;25

So you said,

let's tackle a harder problem.

01;08;08;25 - 01;08;11;29

I think there might be a role

for professional societies.

01;08;12;06 - 01;08;14;23

We haven't had that before in computing.

01;08;14;23 - 01;08;15;02

Right.

01;08;15;02 - 01;08;18;03

So I get to call myself

a computer scientist, and, you know,

01;08;18;03 - 01;08;20;10

and I have some degrees

and some experience,

01;08;20;10 - 01;08;22;04

but I don't have any anything official.

01;08;22;04 - 01;08;26;11

And anybody could just say,

all right, I'm a computer scientist

01;08;26;11 - 01;08;30;13

or I'm a software engineer,

and I'm going to release some software

01;08;30;16 - 01;08;32;24

and they let you do it. That's great.

01;08;32;24 - 01;08;34;29

In other fields, they don't do that.

01;08;34;29 - 01;08;37;08

I couldn't go out tomorrow and say,

you know what

01;08;37;08 - 01;08;40;16

I'm gonna call myself a civil engineer,

and I'm going to go build a bridge

01;08;40;19 - 01;08;41;24

so they don't let you do that.

01;08;41;24 - 01;08;46;06

You need to be certified in order to to,

to do those kinds of things.

01;08;46;09 - 01;08;46;24

And I don't want to

01;08;46;24 - 01;08;50;14

slow down the software industry,

but I think there might be a role to say,

01;08;50;17 - 01;08;54;09

if you get to a certain level

of power of these models,

01;08;54;09 - 01;08;58;00

maybe there should be some certification

of the engineers involved.

01;08;58;03 - 01;08;59;06

mentioned Yann LeCun.

01;08;59;06 - 01;09;02;00

He's really pushing hard

for these open models.

01;09;02;00 - 01;09;04;13

I was saying, you know, wait a minute.

01;09;04;13 - 01;09;07;08

Maybe be good

if somebody is making a query

01;09;07;08 - 01;09;10;22

to do something terrible

that it gets logged somewhere.

01;09;10;25 - 01;09;14;00

And I guess another person I can mention

01;09;14;00 - 01;09;18;04

that I've seen the shift in is,

my colleague Eric Schmidt,

01;09;18;07 - 01;09;20;16

who was, very adamant of saying

01;09;20;16 - 01;09;24;08

we can't have open models

because of the threat from bad actors.

01;09;24;11 - 01;09;26;12

You know, 2 or 3 years ago.

01;09;26;12 - 01;09;29;25

And now he's

he switched and said, it's too late.

01;09;29;28 - 01;09;32;22

These models are powerful enough.

01;09;32;22 - 01;09;35;22

If the bad actors want to use them,

they can create them.

01;09;35;23 - 01;09;38;23

So we might as well harvest

the good of the open models,

01;09;38;23 - 01;09;41;23

because the bad guys have got them

anyways.

01;09;41;23 - 01;09;43;19

And I think that's right. I think.

01;09;43;19 - 01;09;46;03

I think there's nothing

you can do about that now.

01;09;46;03 - 01;09;48;00

AI systems always make mistakes.

01;09;48;00 - 01;09;52;27

Just sometimes it takes

quite a lot of skill to to see them.

01;09;52;27 - 01;09;53;17

Right.

01;09;53;17 - 01;09;56;05

What happens on a even a very functional

AI system?

01;09;56;05 - 01;09;59;12

We still say you will meet the long tail,

find the the outliers,

01;09;59;12 - 01;10;02;12

weirdos, situations

that you did not see coming.

01;10;02;14 - 01;10;07;01

Even when it's highly performance,

expect expect something's going to.

01;10;07;04 - 01;10;08;07

And so

01;10;08;07 - 01;10;11;07

you anticipate

that there will be mistakes.

01;10;11;07 - 01;10;14;16

Now the question is when a mistake touches

a user

01;10;14;19 - 01;10;18;11

who has a particular kind of expectation,

01;10;18;14 - 01;10;20;18

what then happens?

01;10;20;18 - 01;10;23;18

How much how flammable is that

01;10;23;21 - 01;10;26;20

your your AI infrastructure?

01;10;26;20 - 01;10;28;17

Of course, it's all the obvious things.

01;10;28;17 - 01;10;31;17

You have actual infrastructure

in your data pipelines and all the rest.

01;10;31;21 - 01;10;36;08

But it's also things,

intangible things like

01;10;36;11 - 01;10;39;18

at what stage are my user expectations?

01;10;39;21 - 01;10;43;13

Have I managed them sufficiently

where I even could be deploying to users?

01;10;43;13 - 01;10;46;28

What about internally

if I'm making if I'm doing

01;10;46;28 - 01;10;50;26

some, you know, an internal corporate

engineering, I'm offering

01;10;50;26 - 01;10;54;09

some, you know, and I we're looking

at the digital employee experience.

01;10;54;09 - 01;10;58;04

I'm offering some tools to my employees,

some digital tools

01;10;58;07 - 01;11;00;12

have I manage their expectations?

01;11;00;12 - 01;11;02;07

Have I trained my staff?

01;11;02;07 - 01;11;05;21

Do they know how

to think about these tools?

01;11;05;24 - 01;11;06;03

Let's.

01;11;06;03 - 01;11;08;03

I need humans in the loop.

01;11;08;03 - 01;11;10;08

Am I sure my human will be in the loop?

01;11;10;08 - 01;11;12;15

Or might they be asleep at the wheel?

01;11;12;15 - 01;11;13;27

And how do I do the training?

01;11;13;27 - 01;11;16;05

And how do I put in maybe a collection?

01;11;16;05 - 01;11;17;29

Depending on the importance of the task,

01;11;17;29 - 01;11;21;27

I might need to think about

having multiple humans in the loop.

01;11;22;00 - 01;11;24;28

I might need to think about consensus AI.

01;11;24;28 - 01;11;28;29

There are all kinds of,

measurement infrastructure

01;11;29;02 - 01;11;31;28

things that we would need

to put in place in generative AI.

01;11;31;28 - 01;11;35;18

We've just seen endless right

answer this nightmare thing,

01;11;35;18 - 01;11;37;11

a nightmare challenge for management

01;11;37;11 - 01;11;39;09

because we've

all got to change our paradigm,

01;11;39;09 - 01;11;42;09

and we've got to think differently

about measurement metrics.

01;11;42;13 - 01;11;43;06

Have we done that?

01;11;43;06 - 01;11;45;14

Have we put this in place?

01;11;45;14 - 01;11;46;25

Do we have testing pipelines?

01;11;46;25 - 01;11;49;01

Do we have experimentation pipelines?

01;11;49;01 - 01;11;51;28

Do we know how we're going to roll things

back, what we need to do,

01;11;51;28 - 01;11;52;18

we know what we're going

01;11;52;18 - 01;11;56;10

to, well, versus we're going to go,

do we actually know what will happen?

01;11;56;13 - 01;12;00;27

In what kind of scenario do we know

how we're going to make our guardrails?

01;12;01;00 - 01;12;02;05

What sets those guardrails?

01;12;02;05 - 01;12;03;05

How do we update them?

01;12;03;05 - 01;12;06;04

How are we going to react to,

legal changes? Right.

01;12;06;04 - 01;12;10;03

All this stuff now out okay,

I know it's hegemonic

01;12;10;03 - 01;12;14;28

or you can say everything is

AI infrastructure, but to be ready for AI,

01;12;15;01 - 01;12;18;01

there is a lot of stuff

that you would need to be right.

01;12;18;08 - 01;12;23;00

And so one of the ways that you can dodge

a lot of this

01;12;23;03 - 01;12;27;03

is that you

you do outsource some piece to a vendor

01;12;27;06 - 01;12;31;11

who was supposed to do all of it for you,

and you just check that you're getting

01;12;31;11 - 01;12;35;25

precisely what you need, and you have to

still articulate what it is that you need,

01;12;35;28 - 01;12;36;12

and you

01;12;36;12 - 01;12;40;12

have to worry about measurement wise,

that there is going to be a gap or a hole

01;12;40;15 - 01;12;43;09

between what the vendor sees

and what you see.

01;12;43;09 - 01;12;45;26

That's going to be some bit in the middle

that nobody sees,

01;12;45;26 - 01;12;49;17

and that could be a huge risk,

not just in terms of security, but,

01;12;49;20 - 01;12;53;17

in terms of your system

01;12;53;20 - 01;12;57;23

slowly

going, sideways with neither party knows.

01;12;57;26 - 01;13;02;03

Do we need to separate the

AI from the person if we've got, you know,

01;13;02;03 - 01;13;05;07

a cadre of employees who have figured out,

like they've got another tool

01;13;05;07 - 01;13;09;08

in their toolkit,

you know, do we need to care that it's AI?

01;13;09;08 - 01;13;10;20

Is there risk to this?

01;13;10;20 - 01;13;12;20

How should we be thinking about that?

01;13;12;20 - 01;13;16;13

I think this is such a great question

because it goes to the point.

01;13;16;14 - 01;13;18;25

One of the points and I think

sort of one of the underlying things

01;13;18;25 - 01;13;23;01

is, you know, is AI cheating, for example,

you know, yeah, we'll use it.

01;13;23;04 - 01;13;24;23

And I think, no. Right.

01;13;24;23 - 01;13;26;10

I mean, this isn't it's not high school.

01;13;26;10 - 01;13;28;17

You're not getting graded on your,

on your paper. Right?

01;13;28;17 - 01;13;29;05

I have kids

01;13;29;05 - 01;13;31;11

and they can't use generative

AI for writing papers,

01;13;31;11 - 01;13;33;20

but they can use it

for learning biology better.

01;13;33;20 - 01;13;35;24

So it really depends. Right.

01;13;35;24 - 01;13;39;07

But I think that, that most importantly,

first of all, you have to understand

01;13;39;07 - 01;13;42;14

that there are actual, you know, laws

and limitations around this.

01;13;42;14 - 01;13;45;12

Like, you can't just produce something

that's either an imagery or video

01;13;45;12 - 01;13;49;08

or even text, to be honest, and

just put that out into the world of yours

01;13;49;08 - 01;13;50;15

just because you can't copyright it.

01;13;50;15 - 01;13;52;12

That's a that's a legal issue.

01;13;52;12 - 01;13;54;20

But beyond that, we should have everybody

01;13;54;20 - 01;13;58;12

in the organization with guardrails

in place of course using this.

01;13;58;12 - 01;13;59;02

And why is that?

01;13;59;02 - 01;14;02;00

Because it is going to augment

what they are good at.

01;14;02;00 - 01;14;05;00

Sometimes the example uses,

if you put me up against a marker

01;14;05;03 - 01;14;07;26

and you said, okay, in 20 minutes,

both come, both of you

01;14;07;26 - 01;14;11;24

come up with a new idea for a shoe company

or something like that, right?

01;14;11;27 - 01;14;12;05

Yeah.

01;14;12;05 - 01;14;15;11

I would produce something really awesome,

even though I'm not a marketer.

01;14;15;11 - 01;14;18;09

Right. Because ChatGPT would help guide me

and it would be amazing.

01;14;18;09 - 01;14;19;26

And after 20 minutes

it would be incredible.

01;14;19;26 - 01;14;24;14

But the marketer's work product

would be ten times better than mine. Why?

01;14;24;14 - 01;14;27;03

Because they understand

what quality looks like.

01;14;27;03 - 01;14;27;18

The Henderson.

01;14;27;18 - 01;14;29;09

It's sort of like when you say,

hey, write a poem

01;14;29;09 - 01;14;32;06

and you read this poem by ChatGPT,

you're like, this is great,

01;14;32;06 - 01;14;34;25

but a real poet would be like,

that's literal trash.

01;14;34;25 - 01;14;37;06

Like that looks like poem trash, right?

01;14;37;06 - 01;14;41;04

Because the person who actually has

the brain that this, this tools are going

01;14;41;04 - 01;14;45;05

to augment, understand how to guide it,

understand what quality looks like, etc.

01;14;45;05 - 01;14;48;27

and to not have your people using that

as kind of an Iron Man suit,

01;14;49;00 - 01;14;52;01

you're really just shooting

yourself in the foot.

01;14;52;04 - 01;14;53;16

If you work in it.

01;14;53;16 - 01;14;56;16

Infotech research Group is a name

you need to know

01;14;56;19 - 01;14;58;00

no matter what your needs are.

01;14;58;00 - 01;14;59;24

Infotech has you covered.

01;14;59;24 - 01;15;01;28

I strategy covered.

01;15;01;28 - 01;15;04;15

Disaster recovery covered.

01;15;04;15 - 01;15;07;00

Vendor negotiation covered.

01;15;07;00 - 01;15;10;23

Infotech supports you with the best

practice research and a team of analysts

01;15;10;23 - 01;15;14;16

standing by ready to help you

tackle your toughest challenges.

01;15;14;19 - 01;15;19;19

Check it out at the link below and don't

forget to like and subscribe! So.