artificial intelligence is superhuman it

is smarter than you are and there's

something inherently dangerous for the

Dumber party in that relationship you

just can't put the genie back in the

bottle Sam Harris neuroscientist

philosopher author podcaster he goes

into intellectual territory where a few

others their tread six years ago you did

a TED Talk the gains we make in

artificial intelligence could ultimately

destroy us if your objective is to make

Humanity happy and there was a button

placed in front of you and it would end

artificial intelligence what would you

do well I would definitely pause it the

idea that we've lost the moment to

decide whether to hook our most powerful

AI to everything it's just oh it's

already connected to the internet got

millions of people using it and the idea

that these things will stay aligned with

us because we have built them we gave

them a capacity to rewrite their code

there's just no reason to believe that

and I worry about the near-term problem

of what humans do with increasingly

powerful AI how it amplifies

misinformation most of what's online

could soon be fake can we hold a

presidential election 18 months from now

that we recognize as Valor right like is

it safe and it just gets scarier and

scarier I worry we're just going to have

to declare bankruptcy to the internet

if your intuition is correct are you

optimistic about our chances of survival

before this episode starts I have a

small favor to ask from you two months

ago 74 of people that watch this channel

didn't subscribe we're now down to 69 my

goal is 50 so if you've ever liked any

of the videos we've posted if you like

this channel can you do me a quick favor

and hit the Subscribe button it helps

this channel more than you know and the

bigger the channel gets as you've seen

the bigger the guests get thank you and

enjoy this episode

oh

[Music]

son

six years ago you did a TED Talk

um I watched that Ted Talk a few times

over the last week and this Ted Talk was

called can we build AI without losing

control over it in that Ted Talk you

really discussed the idea whether

um AI when it gets to a certain point of

sentience and intelligence will

will wreak havoc on Humanity

six years later

where'd you stand on on it today

do you think are you optimistic about

our chances of

survival survival yeah I mean I can't

say I'm optimistic I'm I am worried

about two

species of problem here that are related

I mean there's sort of the near-term

problem of just what humans do with

increasingly powerful Ai and

um

how it amplifies the the problem of

misinformation and disinformation and

make and just makes it harder and harder

to make sense of reality together

um

and then there's just the the longer

term concern about you know what's

called alignment with with artificial

general intelligence where we build AI

that is is

truly General and you know by definition

superhuman and it's competence and power

and then the question is have we built

it in such a way that is aligned in a in

a durable way with with our interests

and

um

I mean there's some people who just

don't see this problem that they're kind

of blind to it when I'm in the presence

of someone who doesn't have doesn't

share this intuition they don't resonate

to it

I just don't understand what they're

doing or not doing with their minds in

that moment

let's say I'm wrong about that well then

you know it's just the other person's

right and so we're just we just have

fundamentally different intuitions about

about this particular point and then the

point is this

if you're imagining building true

artificial general intelligence that is

superhuman and that is what everyone

whatever their intuitions purports to be

imagining here I mean there's you know

there are people on both sides of the of

the alignment debate

the people who think alignment's a real

problem and or and people think it's a

total fiction

but everyone you know virtually everyone

whose party to this conversation agrees

that we will ultimately build

artificial general intelligence that

will be superhuman and it's in its

capacities

and there's very little you have to

assume to be confident that that we're

going to do that and there's really just

two assumptions one is that intelligence

is substrate independent right there's

no it doesn't have to be made of meat it

can be made in silico right and we've

already proven that with narrow AI I

mean there's just this we obviously have

intelligent machines and you know your

calculator and your phone is better than

you are at arithmetic and it's just

that's that's some very narrow band of

intelligence

so as we keep building intelligent

machines

on the assumption that there's nothing

magical about having a computer made of

meat

the only other thing you have to assume

is that we will keep doing this we will

keep making progress

and eventually we will we will be in the

presence of something more intelligent

than we are

and that's not assuming Moore's Law is

not assuming exponential progress we

just we just have to keep going right

and when you look at the reasons why we

wouldn't keep going those are all just

terrifying right because intelligence is

so valuable and we're so incentivized to

have more of it and every increment of

it is is valuable it's not like it only

gets valuable when you get you know when

you double it or 10x it no no if you

just get three more percent right that's

that's uh that pays for itself

um so

we're going to keep doing this our

failure to do it suggests that something

terrible has happened in the meantime

right we've had a World War we've had a

global pandemic far worse than covid we

got hit by an asteroid something

happened that prevented us as a species

from continuing to make progress in

building intelligent machines right so

absent that we're going to keep going

we will eventually be in the presence of

something smarter than we are

and this is where intuitions divide

my intuition and it's shared by by

many people I'm sure I know at least one

who you've spoken to

my intuition is that

there is something inherently dangerous

for the Dumber party in that

relationship there's something

inherently dangerous for the Dumber

species

to be in Pre in the presence of the

smarter species and we have seen this

you know based on our entanglement with

all other species dumber than we are

right or certainly less competent than

we are

and

buy so buy it reasoning by analogy it

would be true of something smarter than

we are

um

people imagine that because we have

built these machines that is no longer

true right but

and here's where my into tuition goes

from there

that is that imagination is

born of not taking intelligence

seriously right because what

intelligence is is

a you know a mismatching intelligence in

particular is

a

a fundamental

lack of insight into what the smarter

party is doing and why it's doing it and

what it will do next on the part of the

Dumber party right so I mean you just

couldn't imagine that

by analogy just imagine that that dogs

had invented us as their their super

intelligent AIS

right

for the purpose of making their lives

better you know just securing resources

for them securing comfort for them

making getting the medical attention

um

it's been working out pretty well for

the dogs for about 10 000 years right I

mean there's some exceptions we've got

we've mistreat certain dogs but

generally speaking

for most dogs most of the time humans

have been a great invention right now

it's true

that

the

the mismatch in our intelligence

dictates a fundamental blindness with

respect to what we've become in the

meantime right so like we have all these

instrumental goals and things we care

about that they cannot possibly conceive

right they know that when we go get the

leash and say it's time for a walk they

understand that particular part of the

language game but everything else we do

when we're talking to each other when

we're on our computers or on our phones

they don't have the dimmest idea of what

we're up to and

if we ever if something happened if we I

mean we love the truth is we love our

dogs we make just irrational sacrifices

for our dogs we prioritize their health

over all kinds of things

that is just amazing to consider

and yet

if we learn if there was a new you know

Global pandemic kicking off and some

xenovirus was jumping from dogs to

humans and it was just a super Ebola

right it was just it was 90 lethal and

if this was just a forced choice between

him what do you value more you're the

the lives of your dogs or the lives of

your kids right if that's if that's the

situation we were in it's totally

conceivable unless it's not a you know

not by no means impossible

we would just kill all the dogs right

and they would never know why right we

would just and it's because we have this

layer of of

mind and culture and and just just the

the new sphere right there's just this

this realm of of of mind that requires a

requisite level of intelligence to even

be party to even know exists

that they have they have no idea it

exists right and it's

so this is a fanciful uh analogy because

the dogs did not invent us but evolution

invented us right Evolution has coded us

you know as I said to survive and spawn

and that's it right so Evolution can't

see

everything else we've done with our time

and attention and and all the values

we've formed in the meantime and all the

ways in which

we have explicitly disavowed the program

we've been given right so Evolution gave

us a program

but if we were really going to live by

the likes of that program

what would we be doing and we would be

having as many kids as possible right

you know the guys would be going to

sperm banks and donating their sperm and

finding that like the best use of their

time and attention it's like the idea

that you could have hundreds of kids

for which you have no financial

responsibility

that would be the that should be the

most rewarding thing that you could

possibly do with your time

as a man

and

yet that's obviously not what we do and

there are people who decide not to have

kids and they're people who and and yet

and everything else we do

from you having podcast conversations

like this to to

curing diseases type of Interest like

literally everything we're doing with

our you know with science with with

culture

is yes there are points of contact but

between those those products and our

evolved capacities right it's not it's

not just it's not magic right we are

social primates that that have leveraged

certain ancient Hardware to do new

things

but evolution the code that we've been

given doesn't see any of that right and

we've not been optimized to build

democracies right

um Evolution knows nothing it can know

nothing if evolution were a coder

there's just no there's no democracy

maximization in that code right it's

just it's not a it's not there so the

idea that these things

will stay aligned with us because we

have built them because if we have this

origin story that we gave them their

initial code and yet we gave them a

capacity to rewrite their code and build

future generations of themselves right

um

there's just no reason to believe that I

see no and and the mismatch

in intelligence is intrinsically

dangerous and you could see this by

I mean it's just Stuart Russell I don't

know if you had him on the podcast he's

a great

um

professor of computer science at

Berkeley and he he wrote literally

co-wrote one of the most popular

textbooks on AI

um

he has some arresting analogies which I

think are good intuition pumps here um

and one is just think of how you would

feel if

you knew like unless we got a

communication from elsewhere in the

galaxy

and it was a message that we decoded and

it said people of Earth we will arrive

on your lowly planet in 50 years get

ready

right

that anyone who thinks that we're going

to get super intelligent AI in let's say

50 years

things were were essentially in that

situation

and yet we're not responding emotionally

to it in the same way if we if we

received a communication from a a

species that we knew

just by by fact by the sheer fact that

they were communicating with us in this

way we knew they were more competent and

more powerful and more intelligent than

we are right and they're going to arrive

right

we would we would feel

that we were on the threshold of the

most momentous change in the history of

of our species

and we would feel but most importantly

we would feel that it's because

this is a a relationship an unavoidable

relationship that's being foisted upon

us right it's like we like some a new

creature is coming into the room

right with its own capacities

and now you're in relationship and and

one thing is absolutely certain it is

smarter than you are right by by what

factor I mean ultimately we're talking

about by factors you know just by so

many orders of magnitude it's it

our intuitions completely fail I mean

even if even if it was just

a difference in in the time of

processing even if it let's say there

was no difference in in the actual you

know native intelligence but it's just

processing speed

a million fold difference in processing

speed is is just

a phantasmagorical difference in

capacity so like just imagine we had 10

Smart Guys in a room over there and they

were working and thinking and talking a

million times faster than we are right

well so they're no smarter than we are

but they're just faster

and we talked to them once every two

weeks just to catch up on you know what

they're up to and what they want to do

and whether they still want to

collaborate with us well

two weeks for us is 20 000 years of

analogous progress for them

right so how could you how could we

possibly hope to constrain the opinions

and collaborate with and negotiate with

people no smarter than ourselves who are

making twenty thousand years of progress

every time we make two weeks of progress

right it's just it's it's

is unimaginable and yet

there are many people who don't that

just think this is just fiction

everything I all all the noises I've

made in the last five minutes are just

like a

a new religion of fear right and it's

just there's no reason to think that

alignment is even

a potential problem if your intuition is

correct and the analogy of us getting a

signal from outer space that someone is

coming in 30 years which by the way a

lot of people that speak on this subject

matter

um don't believe it's even going to be

30 years yeah until we reach that sort

of Singularity moment I think they speak

of artificial general intelligence I've

heard people like Elon say you know

many fewer decades 10 10 years 15 years

20 years Etc if that is correct then

surely this is the most pressing

challenge conversation issue of our time

and there's no logical reason

that I can see to

refute your intuition there

I I can't see a logical reason the rate

of progress will continue

don't necessarily see anything that will

wipe out or pause our rate of progress

um

I mean let me just to uh be charitable

to the other side here there are other

assumptions that they smuggle in that

they some people I mean some do it

without

being aware of it but some actually

believe these assumptions and This

spells the difference on on this on this

uh particular intuition

um so so it's possible to assume that

the more intelligent you get the more

ethical you become by definition right

now and we might you know draw a

somewhat more equivocal picture from

just the human case where we see that oh

there's some very smart people who

aren't that ethical but

they're I believe there are people I

mean I've talked to a few at least a few

people who believe this

there are people who assume that kind of

in the limit as you push out into just

just

if

far beyond human levels of intelligence

there's every reason to believe that all

of the the

provincial creaturely

failures of human ethics will be left

behind as well it's like you're not like

the the selfishness and the and the and

the basis for conflict like what like

these are not the apish urges of you

know status seeking

uh

monkeys is is just not

it's not going to be in the code and as

you push out into into just kind of the

Omnibus Genius of of the coming AI

you're gonna there's a kind of a

sainthood that's going to come along

with it right and and a wisdom that will

come along with it now

I just think that's a that's quite a

gamble I I think I would take the other

the other side of that bet and and I

would frame it this way

there have to be ways in the space of

all possible

intelligences that are beyond the human

right there's got to be more than one

possible there's got to be just like

there's many different ways to have a

chess engine that's better than I am at

chess

they're still they're they're different

from each other but they're all better

than me right

um there's got to be more than one way

to have a superhuman artificial

intelligence

and I would I would imagine there there

are

you know not not an infinite number of

ways but just a vast number of of

in the space of all possible Minds there

are many locations in that space beyond

the human

that are not aligned with human

well-being right there's got to be more

ways to build this

unaligned then aligned right and what

other people are smuggling into this

conversation is the intuition that no no

once you get beyond the human

is just going to get it's just you're

going to be in the presence of you know

just the Buddha who understands quantum

mechanics and oncology and everything

else right

I just see no reason to think that

that's so and we we could build

something that is

again taken intelligence seriously

we're going to build something we're in

relationship to it's really intelligent

in all the ways that we're intelligent

it's just better at all of those things

than we are it's by definition

superhuman because the only way it

wouldn't be superhuman the only way it

would be human level even for 15 minutes

is if we didn't let it improve itself if

we wanted to just keep it stuck at you

know at a we just we built a college

undergraduate we wanted just to keep it

stuck there but we would have to dumb

down all of the specific capacities

we've already built right just like all

every AI we have narrow AI is superhuman

for the thing it does you know it's it's

it has access to all the information on

the internet right it's just like it's

got perfect memories it can perfectly

copy itself when one part of the system

learns something the rest of the system

learns it because it just can swap files

right it can it's um you're again your

your phone is a bit is a superhuman

calculator there's no reason to make it

a a calculator that is human level

um

and so we're never going to do that

we're never going to be in the presence

of human AGI we will be immediately in

the presence of superhuman AGI and then

the question is how quickly it it

improves and how far they're how much

Headroom is there to improve into

on the assumption that you can get quite

a bit more intelligent than we are right

that there's like they were nowhere near

the summit of possible intelligence

you have to imagine that you're going to

be in the presence of something

that is again it could be completely

unconscious right this I'm not saying

that there's something that's like to be

this thing

although there might be and that's a

totally different problem that's worth

worrying about

but

whether conscious or not

it is solving problems detecting

problems

improving its capacity to do all of that

in ways that we can't possibly

understand

and the products of its increasing

competence are always being surfaced

right so it's like it's

we've been we've been using it to change

the world we became we've become Reliant

upon it we built this thing for a reason

I mean one thing that's been amazing

about the developments in recent months

is that

those of us who have been at all

cognizant of the AI safety space for you

know now going on a decade or more for

some people

always assumed

that as we got closer to the end zone

we'd become either the labs would become

more circumspect we'd be building this

stuff

air gap from the internet you know it's

like we have this phrase air gapped from

the internet like we thought this was a

thing like you this thing would be in a

box and then the question would be

well do we let it out of the box and let

it do something right like is it safe

and how do we know if it's safe right

and we thought we would have that moment

we thought it would it would happen in a

lab at Google or at Facebook or

somewhere we thought we would hear okay

we've got something really impressive

and now we just want it to touch the

stock market or we wanted to touch the

our medical data or we just want to see

if we can use it

we're way past that right we've built

this stuff already in the wild it's

already connected to the internet it's

already got millions of people using it

it already has apis it's already it's

already doing work so from an AI safety

point of view that's amazing like we

didn't even have the moment the the

choice point we thought was going to be

so

fraught of course we didn't we we

because there was such

pressing incentives for people to press

forward regardless of that conversation

especially but yeah everybody everyone

everyone thought I mean I was never I

was I don't believe I was ever in

conversation with someone with someone

like Elias or udakowski or or Nick

Bostrom or Stuart Russell

who assumed we would be in this spot

like I just everyone we because

you know I'd have to go back and look at

those conversations but

there was so much time spent you know it

seems quite unnecessarily on this idea

that

circumspect we'd make a certain amount

of progress and circumspection would

kick in like even the people who are who

were doubters would become worried and

at their and there would be like in the

final yards you know as we go across

into the end zone

there'd be some mode where we could sort

of slow down and figure it out and try

like try to deal with the arms race

Dynamics like let's place a phone call

to China and and and just like let's

talk about this we got something

interesting but the stuff has already

been built in connection to everything

and there's already just endless

businesses being

being

devised on the on the the back of this

thing and all the improvements are going

to get plowed into it and so just

imagine what this looks like even in

success right like let's say it just

starts working wonders for us and we

just we get these great productivity

gains and

okay

so then we cross into the into the you

know whatever the singularity is right

at whatever speed we find ourselves in

the presence of something that is truly

General after all of this stuff is all

of this narrow stuff

uh albeit superhuman narrow stuff is

is something that we totally depend on

right like every hospital requires it

and every airplane requires it and all

of our missile systems require it and

it's we're just this is the way we do

business

um

there is no there's nothing to turn off

at that point I mean I just don't you

know it's like I guess

I mean I put this to Mark Andreessen on

my podcast and he said yeah you can turn

off the internet I mean I don't I can't

believe he was quite serious I mean yes

if you're North Korea I guess you can

turn off the internet for North Korea

and that's why North Korea is like North

Korea but the idea that

we could I mean it just the cost of

turning off the internet now would be

uh

I think it would be unimaginable in the

in the in the economic just the economic

cost alone

it just would be

um

so anyway I'm interested the the idea

that we've we've lost the moment to

decide whether to hook our most powerful

AI to everything

because it's already being built more or

less in contact with if not everything

many so many things that you just can't

put the genie back in the bottle that's

that is genuinely surprising to me and

um

yeah I mean incentives is this not the

most pressing problem then because I I

was going to ask about this conversation

by asking you the question about the

thing that occupies your mind the most

and the most important thing we should

be talking about and I I in part assume

the answer would be artificial

intelligence because the way that you

talk about your intuition on this

subject matter you've got children

yeah you think about the future a lot

um if you can see this

species coming to Earth in the next even

if it's in the next 100 years

um it strikes me to be the most pressing

problem for Humanity

well I do I'm

as interesting as I think that problem

is and and consequential as it is

I'm

I'm worried that life could become

unlivable in the near term before we

even get there like I'm just worried

about the the misuses of narrow AI in

the meantime just I'm worried about just

just take the current level of AI we

have you know we have gpt4

um

I I think within the next 12 months or

two years let's say let's say we

whatever GPT 5 is

we're going to be in the presence of

something where

most of what's online that purports to

the information could soon be fake right

we're like

just most of the text do you find on any

topic is just fake right like someone

has just decided write me a thousand

Journal articles on why mRNA vaccines

cause cancer

and give me you know 150 citations write

them in the in the style of Nature and

nature genetics and Lancet and Jama

um and just put them out there right

right one teenager could do that in five

minutes with the right AI right it's

like it's just like we're not gpd4 is

not quite that but

gpt5 you know possibly will be them it's

like that that is such a near-term

advance right or get you know just when

you imagine knitting together

the visual stuff like mid-journey and

Dolly

um

and stable diffusion with with a large

language model

just imagine the tool again this is

maybe this is 18 months away maybe it's

three years away but it's not 30 years

away

the tool which where you can just say

give me a 45 minute documentary on how

the Holocaust never happened filled with

archival imagery give me you know Hitler

speaking in German and with it with the

appropriate translations and

um

give it give it in the style of Alex

Gibney or Ken Burns or and

give me a 10 000 of those right like

that like that's all all the friction

for misinformation has been taken out of

the system and

yeah I worry we're just going to have to

declare bankruptcy with respect to the

internet like just like we just are not

going to be able to figure out

what's real and when you when you look

at how hard that is now with social

media uh in the in the aftermath of of

covid and Trump

and how just the challenge for of

holding an election that most of the

population agrees was valid right that

challenge

already

is is

on the verge of being insurmountable in

the U.S right

I mean it's just like it's easy to see

us failing at that AI aside now when you

add a large language models to that and

the more competent

future version of it where it's just the

most compelling deep fakes are

indistinguishable from

you know real data

um

and everyone is siled into their tribes

where they're stigmatizing the

information that comes from any other

tribe and we're just and the internet is

now so big a place that

there really isn't the ordinary

selection pressures where where bad

information gets successfully debunked

so that it goes away it says you can

live in a conspiracy cult

for the rest of your life if you want to

you know you can be queuing on all day

long if you want to

and now we've got deep fakes Shoring all

that up

and just spurious you know scientific

articles showing all that up I all this

becomes a more compelling form of

psychosis and you know culturally

speaking and so I'm just worried that

it's good it's going to get harder and

harder

for us to cooperate with one another and

collaborate

and that our politics Will just

completely break and that'll you know

offer

an opportunity for lots of

you know Bad actors and I mean leaving

aside there's cyber terrorism and

there's their synthetic biology that you

know the moment you get you turn AI

loose on on the on the prospect of

of engineering viruses and you know all

of that it's like it it potentiates I

mean the asymmetry here is that

it seems like it's it's always easier to

break things than to fix them or to

prevent people categorically prevent

people from breaking them and what we

have with increasingly powerful

technology is the ability for one person

to create more and more damage or one

small group of people right and it was

so it's just

it just turns out it's hard enough to

build a nuclear bomb that like one

person can't really do it you know no

matter how smart you need a team and you

need to you need

it's traditionally you've needed State

actors and you need you need access to

resources and you have to get the

physical material and

it's hard enough but this isn't this is

being fully democratized this Tech and

so it's

um

yeah I worry about the near-term chaos

I've never found the narrow term

consequences of artificial intelligence

to be that interesting until now is that

what you said that image of like the

internet becoming unusable so that was a

real Eureka moment for me because I've

not been thinking about that yeah no me

too I was I was just concerned about the

AGI risk and now

really in the in the aftermath of trump

and kovid I've just I see the risk of

um

you know it if not losing everything

losing a lot that matters

just based on our

and interacting with just these very

simple

tools that that are Mis reliably

misleading us I mean I'm just I'm amazed

at what social media

I forget about I'm amazed at what

Twitter did to me I mean you know even

with all of my

training and all you know with my head

screwed on reasonably straight I mean

it's amazing to say it but

almost all of the truly bad things that

have happened to me in the last decade

that just really like just destabilized

relationships and

and just priorities and really like I

kind of got plowed back into kind of

became a kind of professional emergency

you know stuff I had to respond to you

know in writing or on podcasts

it was on Twitter it was my my

engagement with Twitter was the thing

that produced the chaos and it was

completely unnecessary

um

and it was just it was amplifying a kind

of signal for me that I felt compelled

to pay attention to

because I was on it and I was trying to

communicate with people on it I was

getting certain communication back and

it was giving me a picture of the rest

of humanity which I now think was

fundamentally misleading but it was it

was still consequential in its yeah like

even believing it was a certain point

believing that it was misleading wasn't

enough to inoculate me against the

delusion of these kind of the opinion

change that was being forced upon me

um and I was feeling like okay like

these people are becoming unrecognizable

like I know some of these people I've

had dinner with some of these people and

their behavior on Twitter is is

appearing so deranged to me and so in

such bad faith

um

the people are uh people who I know to

be

non-psychopaths are starting to behave

like Psychopaths at least on Twitter and

I'm becoming similarly unrecognizable to

them that it's just again it it all felt

like a psychological experiment to which

I hadn't consented which I enrolled

myself somehow because it was

it was what everyone was doing in 2009

um and I spent you know 12 years there

getting some signal and responding to it

and

it's not to say that it was all bad I

mean I read a bunch of good articles

that got linked there and I you know I

discovered some interesting people but

uh

the change in my life after I deleted my

Twitter account was so enormous I mean

it's embarrassing to admit it I mean

it's just it's like it's like getting

out of a bad relationship and it was

just it was a fundamental

um

just freedom from from this this chaos

Monster that was it was always there

ready to disrupt something but based on

its own Dynamics and when did you do

Lisa

um yeah like December I think it was

December I would and I'm not someone

that really takes sides on things I like

to try and remain in the middle I think

politically so you must have a very

different Twitter experience than I was

having no no no

so I don't treat anything other than

this podcast trailer don't do anything

else right okay so I just did anything

you'll see on my Twitter is the podcast

trailer that's it yeah and for all the

reasons you've described and more

interestingly I wanted to say in the

last eight months as someone that tries

to be doesn't get caught up too much in

the media oh Elon bought this it's a

hundred percent gone in that direction

as in my timeline now is I say to my

friends all the time and some of my

friends who are again I think are

nuanced and balanced have said to me the

there's something that's been turned up

in the algorithm to increase engagement

that has planted me in an unpleasant

Echo chamber that I didn't desire to be

in and if I wasn't Cog somewhat

conscious I would 100 be in there my

timeline my friend tweet the other day

my friend Castle tweeted he's never seen

more people die on his Twitter timeline

than he has in the last six months

they're prioritizing video so you're

seeing a lot of like

death in CCTV footage that I've never

seen before and then the debate around

gender

um politics

right-leaning subject matter has never

been more right down your throat yeah

because it's been it's almost like

something in the algorithm has been

switched where it's now it's now like

people have been let out the Asylum

that's you know I can describe it

and it's made me retract even more so

when Zuckerberg announced threads the

other the other couple of weeks ago

it was kind of like a a life raft right

out of this out of the Titanic

um and I really really mean that and I'm

not someone to get easily caught up in

narrative you know as it relates to

social media platforms it's been my

industry for a decade but what I've seen

on Twitter and it's actually made me

believe this hypothesis I had five years

ago where I thought they would be

um I thought the the Journey of social

networking would be would have way more

social networks and they'd be more siled

I thought we'd have one for our

neighborhood our football club and now I

believe that even more than ever yeah

that seems right and I think I mean

whether it's possible to have a truly

healthy social network that people want

to be in and it's a good reason to be

there and it's it's uh I don't know if

that's possible I I like to think it is

but it's um

I think there's certain things you you

have to

clean up

at the outset that is supposed to make

it possible I mean I think I think

anonymity is a bad thing I think

um

probably being free is a bad thing I

think for you if you know you sort of

get what you pay for online and if it's

if it's uh I just think they're there

there might be ways to set it up that

where it would be better but I don't

think it would be popular

was that I think with the thing that

makes it popular makes it toxic right

right and even the anonymity piece I've

played this out a couple of times in my

mind and the rebuttal I always get is

well there's people in Syria who have

news to break important needs to break

and they they'd be hung if they

so we need a Anonymous version of the

social internet right yeah well I guess

there could be some exception there but

um

I don't know it just doesn't it actually

doesn't interest me

because

I just feel such

a different sense of

my being in the world as a result of not

paying attention to

the my online

simulacrum of myself is it's it's a um

because Twitter was the only one I used

like I was on I've been on Facebook this

whole time I've been on I think I I

guess I'm on Instagram too but like my

team just uses those as marketing

channels you know it's just like you it

sounds like that's the way you use

Twitter now but Twitter was the the one

that I decided okay this is going to be

me

I'm gonna be posting here I'm gonna you

know if if I've made a mistake I want to

hear about it you know it's like and I

just want to use it as actual uh an

actual basis for communication

um

and for the longest time it actually

felt like a valid tool in that respect

you know it reached a crisis point I

decided this is just pure toxicity

there's just no reason even the good

stuff can't possibly make a dent in the

bad stuff so I just deleted it and then

I was I was returned to the real world

right where I've where I actually live

and to books and to I mean I'm online

all the time anyway but up but it's not

having the this is the time course of

reactivity

when you don't have social media when

you don't wait and you don't have a

place to put this this instantaneous hot

take

that you're tempted to be put out into

the world because there's literally no

place to put it like if like for for me

if I have some reaction to something in

the news

I have to decide whether it's worth

talking about it in my next podcast that

I might be recording you know four days

from now

and rather often people have been just

bloviating about this thing for four

solid days before I ever get to the

microphone and I then I get to think

well this is still worth talking about

in most almost nothing survives that

test anymore right it's like the

conversations moved on

so there's actually no place for me to

just type this thing that either takes

me 10 seconds and then rolls out there

to get

to detonate in the minds of you know my

friends and enemies to opposite effect

and then I see the the result of all

that you know on a again on a the sort

of reinforcement Loop of every 15

minutes

um

not having that is such a relief that I

just don't even know why I would so like

when threads was announced I wasn't I

think I'm on threads too but it's not me

it's just you know just again another

marketing channel

um but yeah I haven't I feel such relief

not

exercising that muscle anymore where

it's like I I you know I don't know how

often I was checking

Twitter but it was I was you know I was

not checking it just to see what was

happening to me or what the response to

my La the last thing I tweeted I was

checking it a lot because it was my news

feed it's like I'm following you know

200 smart people they're telling me what

they're paying attention to and so I'm

fascinated so yeah well yeah I want to

see that next article or that next video

just that engagement and the endless

opportunity to comment and to put my

foot in my mouth or put my foot in

someone else's mouth or have someone put

their foot it's just

not having that has been such relief

that I would be I mean it's not

impossible but I would be very

cautious in reactivating that because it

was it was so much noise and again it

would it created

there's so much it became a

uh I mean it became an opportunity cost

but it became a just this endless

opportunity for misunderstanding but

especially misunderstanding of me and

you know everything I've been putting

out into the world and then my sense

that I had to react to it and then you

just can't plow that

back into the

you know that that becomes the basis for

further misunderstanding

um

and it just it constantly was giving me

the sense that there's something there's

something I need to react to on my

podcast in an article on Twitter

that

it's just this is a valid signal like

this is this is this is like this is a

five alarm fire this is like you got to

stop everything like you're by the pool

on the One Vacation you're taking with

your family that summer and

this thing just happened on your phone

that like it can't wait right like you

actually have to pay attention because

it's like the conversation is happening

right now and so it was a kind of

addiction to information and right you

know some level reputation management or

or or

[Music]

um

and it was just I mean just yeah to just

be free of it is such a relief apart

from like you know

health issues with certain family

members

virtually the only bad things that have

happened to me have been a result of my

engagement with Twitter over

the last 10 years

so it's just it's just you know I I you

know I guess I'm if I'm a masochist I

would be back on Twitter but like that

would be the only reason to do it narrow

AI I asked you the question a second

year which we um

I really wanted to get a solution to it

because I'm mildly terrified I

completely believe you'll believe your

um the logic underneath your opinion

that narrow area will cause this

um destabilization and usability of the

internet

so just focusing on narrow AI what what

would you consider to be a solution to

prevent us getting to that world where

misinformation is right to the point

that it can destabilize Society politics

and culture

well I think it's something I've been

asking people about on my podcast is

because it's not actually my wheelhouse

and I would just need to hear from

experts about what's possible

technically here but

um

I'm imagining that

paradoxically or ironically

this could Usher in a new kind of

gatekeeping that we're going to rely on

because like the provenance of

information is going to be so important

I mean the the the the assurance that a

video has not been manipulated or

there's not a a just a pure confection

of

of deep fakery

right so you get so

it could be that we're we're Meandering

into a new period where

you're not going to trust a photo unless

it's come it's coming from you know

Getty Images or you know the New York

Times has some story how the about how

they have verified every photo in their

that they put in their newspaper they

have a process and you know so if you

see a a a video of of Vladimir Putin

seeming to say that he's

declaring war on the U.S

right I think

most people are going to assume that's

fake until proven otherwise it's like

it's just it's just going to be too much

fake stuff and it's going to be it's all

going to look so good that the New York

Times and every other you know

organ of media that we have relied upon

um as imperfect as they've been of late

they're going to have to figure out what

the tools are whereby they can say okay

this is actually a video of Putin right

and if the new I mean I'm not going to

be able to figure it out on my own right

the New York Times doesn't have a

process or CNN doesn't have a process

that they go through before they say

Okay Putin really said this and so this

is we have to now react to this because

this is real

um

whatever that process is and you know

whether it's whether there's some kind

of digital Watermark that you know

that's connected to the blockchain

that's I mean there's there's some tech

implementation of it that can be fully

democratized where you by just being in

the latest version of the Chrome browser

can know that you're so you you can

differentiate you know real and fake

videos say I don't know what the

implementation will be but I just I just

know we're going to get to some spot

where it's going to be all right

we have to declare epistemological

bankruptcy we don't know what's real we

have to assume anything especially lurid

or agitating is fake until proven

otherwise so prove otherwise

and that's you know that that'll be a

resetting of something I don't know

what we do with that in a world where we

really don't have that much time to

react to certain things that are you

know a video of Putin saying he's

launched his big missiles

is something that you know 30 minutes

from now we would we would understand

whether it's real or not

I mean forget about again forget about

everything we just said about AI

look at all of our Legacy risks look at

the risk of nuclear war the the risk of

stumbling into a nuclear war by accident

has been hanging over our head for 70

years I mean we've got this old Tech

we've got these wonky radar systems that

throw up errors we've we have

moments in history where

you know one Soviet sub-commander

decided based on his just gut feeling

his common sense that the data was

almost certainly an error and he decided

not to pass the the the the obvious

evidence of a an American ICBM launch up

the chain of command knowing that the

chain of command would say okay you have

to fire

right and he reasoned that if the U.S

was going to attack the Soviet Union

they would launch more than I think in

this case it looked like there were four

missiles that was the radar signature

if the us is going to launch a first

strike against the the Soviet Union in

one of this like the mid 80s

um

they're going to launch more than four

missiles right this has to be this has

to be bad data right so this is that but

you know so if we automate all this will

we automate it to systems that have that

kind of common sense right

um

but

we've been perched on the on the edge of

the Abyss based on this this the

possible forget about malevolent actors

you know who might decide to have a

nuclear war on purpose we have the

possibility of of accidental nuclear war

you add this cacophony of misinformation

and deep fake to all of that and it just

gets scarier and scarier and this is

just this is not even AI this is just

you know you know narrow AI Amplified

misinformation how do you feel about it

well I mean this is the thing that

worries me I I worry about the next

election you know I think the next

president if we can run the 2024

election in a way that most of America

acknowledges was valid

that will be an amazing Victory you know

whatever the outcome I mean obviously I

would not be looking forward to a trump

presidency but um

I think even more fundamental than that

is can we hold a presidential election

18 months from now

that is that we recognize as valid right

like that I I don't know I don't know

what kind of resources are being spent

on on that particular

performance but that is hugely important

and

I don't think

our near-term experiments with AI is

going to make that easier why is it so

important

well it's just I mean if you think the

maintenance of

of uh a valid democracy in in the

world's low in superpower is is of minor

importance I

um I'd like to drink the tea you're

drinking

but yeah it's Mystic

I mean I I'm I can't say I'm optimistic

I'm you know it's

it's a paradoxical State I mean because

I I definitely I I tend to focus on

what's wrong or might be wrong

I tend to

I think have a a pessimistic bias right

like I I tend to notice what's wrong as

opposed to what's right you know I mean

that's my

um

that's my bias but

I'm actually very happy right like I

have a very a very good life I'm just

like everything is is I just I'm

incredibly lucky I'm surrounded by great

people it's like it's just it's all

great

and yet I see all of these

risks on the horizon so I'm like I'm not

um

I just I have a very high degree of

well-being at this moment in my life and

yet I like what's on the television is

scary and so it's it's it's very

interesting juxtaposition yeah

you know I will be I'll be very relieved

if we have a

busy or just I feel like we're in a very

weird spot I mean like the I haven't

seen a

a full postmortem on the coveted

pandemic that has fully encapsulated

what I think we what I think happened to

us there but my my vague sense is that

we didn't learn a whole hell of a lot I

mean basically what we learned is we're

really bad at responding to this kind of

thing this was a challenge that that

just fragmented us as a society it could

have brought us together

it didn't

and

it it Amplified all of the the divisions

in our society politically and and

economically and tribally in all kinds

of ways the role of misinformation and

disinformation on all of that was was

all too clear and I think just getting

worse so I think you know as a dress

rehearsal for some future pandemic

that's that is inevitably going to come

and is you know could well be worse I

think we failed this dress rehearsal and

you know I have to hope that at some

point our institutions

will reconstitute themselves so as to be

obviously trustworthy and engender the

kind of trust we actually need to have

at our institutions like we need a CDC

that not only that we trust but that is

trustworthy that we that we that we're

right to trust right

and and so it is with an FDA and every

other you know institution that that is

relevant here and

we don't quite have that and half of our

society thinks we don't have that at all

right and and so it's

um we have to rebuild trust in

institutions somehow and I just think

you know we have a lot of work to do but

to even figure out how to make

an increment of progress on that score

because

we're again the siloing of of large

constituents into alternate

information universes is just just not

functional and that's so much of what

social media has done to us and

alternative media I mean like you know I

call it you know you and I are

podcasters but I call it podcast to Stan

right we have this this landscape of I

mean there's now whatever million plus

podcasts and there's

you know you email newsletters and

everyone has now just decided to curate

their information diet in a way that's

just bespoke to them and

you can stay there forever and you're

getting you're getting one slice of and

it could be a you know

a completely fictional slice of of

reality and

um we're losing the ability to converge

on a common picture of what's going on

and you

so did that sound optimistic I didn't

hear the optimism there you tell me no I

I no I but I I kind of can't refuse to

anything you said on a like a logical

basis it all sounds

um like that is the direction of travel

that we're going in unfortunately

um I have faith that they'll be

surprising

positives it always tends to be

surprising positives that we also didn't

factor in

um it's easy to see

I mean if there's anything if there's

any significant low-hanging fruit

technologically or or

scientifically that could be AI

enabled for us let me just take like you

know a cure for cancer a cure for

Alzheimer's I mean just having one thing

like that

right that would be such an enormous

good

um

and that so that that is that's what

that's why we can't get off this ride

and that's why there is no break to pull

because the value of intelligence

is so enormous I mean it is it is just

it's not everything I mean it's not that

you know there's there are other things

we care about and a right to care about

Beyond intelligence I mean love is not

the same thing as intelligence right but

intelligence is the thing that can

Safeguard everything you love right like

even if you think the whole point in

life is to just get on a beach with your

friends and your family and just hang

out and enjoy the sunset

okay

you don't have to augment you you don't

need superhuman intelligence to do any

of that right you're you're fit to do it

exactly as you are you could have done

that in the 70s and it would just be

just as good a beach and they'd be just

as good friends

but

every gain we make in intelligence is

the thing that safeguards that

opportunity for you and everyone else

how would you I feel like we've not

defined the time artificial general

intelligence from my understanding of it

it's when the the intelligence can think

and make decisions almost like a human

yeah maybe Loosely this

this is a kind of just a semantic

problem but

intelligence can mean many things but

you know Loosely speaking it is the

ability to solve problems

uh and

meet goals make decisions

in response to a changing environment in

response to data

um

and the general

aspect of that is an ability to do that

in across in many different situations

all the sort of situations we encounter

as people

and to have one's capacity in one area

not you know as I get better at deciding

whether or not this is a cup I don't

magically get worse at deciding whether

you know you just said a word right it's

like I can do but it's like I can do

multiple things in multiple channels

that's not something we had in our

artificial systems for the longest time

because we were everything was bespoke

to the task we'd build a chess engine

and it couldn't even play Tic-Tac-Toe

all I could do was play chess and they

could and we and we just would get

better and better in these in these

piecemeal narrow ways and then

things began to change a few years ago

where you'd get you know with like deep

mind would it would have its algorithms

that were

uh you know the same algorithm with

slightly different tuning could play go

right or it could you know it could

solve a protein folding problem as

opposed to just playing chess right and

it became the best in the world at chess

and I became the best in the world to go

and

um

and amazingly I mean to take you know

Alphas what Alpha zero did

it

you know before Alpha zero

all the chess algorithms were they just

had all of our chess knowledge plowed

into them if they had studied every

human game of chess and they just it was

just you know it was

it was a bespoke chess engine Alpha zero

just played itself I think for like four

hours right it just it just had the

rules of Chess and then it played itself

and it became better not

merely than every other every person

who's ever played the game it became

better than all the chess engines that

had all of the the all of our chest

knowledge plowed into them so you it's a

fundamentally new moment in in how you

build an intelligent system

and it promises this this possibility

again

this inevitability the moment you admit

that we will eventually get there the

moment the moment you admit

that it's it can be done in silico

and the moment that you admit that we

will just keep going unless a

catastrophe happens

and those two things are so easy to

admit that I just don't at this point I

don't see any place to stand where

you're not forced to admit them right I

don't see any neuroscientific or

cognitive scientific

argument for

substrate dependence for intelligence

given what we've already built and

again we're we're going to keep going

until something stops us right we'll hit

some immovable object that prevents us

from releasing the next iPhone but other

otherwise we're going to keep going and

then yeah so then it then we'll whatever

General will mean

in that first case they'll be a case

where

we've built a system that is so good at

everything we care about

that is functionally General now maybe

it's missing something maybe it's not

you know maybe it's missing something

that we don't even have a name for you

know we're missing all kinds of their

possible intelligences that we haven't

even thought about because we just

haven't thought about them right there

there's the things that there are ways

to section the universe undoubtedly that

we can't even conceive of because we are

just we have the minds we have Elon was

asked a question on this by journalists

the journalist said to him in a world

where you believe that to be true that

artificial general intelligence is

around the corner when your kids come to

you and say Daddy what should I do with

my life

Define purpose and meaning

what advice do you now give them if you

hold that intuition to be true

that it's around the corner

what do you say to your children when

they say what should I do with my life

to create purpose and meaning and

did you say that Elon answered this

question yeah what did he say it's one

of the most chilling moments in an

interview I think I've seen in recent

times because he stutters he goes silent

for about 15 seconds which is very

annelon he stutters he stutters

um he stutters a bit more because he

can't

and then he says he thinks he's suspend

he's living in suspended disbelief

because if you've really thought about

it too much what's the point he says

what's the point of me building all

these cars he was in his Tesla Factory

what's the point of me building all

these cars and what's the point I do

think that sometimes so I think I have

to live in as his words were suspended

disbelief right well I would encourage

him to ask what's the point of spending

so much time on Twitter because that he

could clearly benefit from

rethinking that but

um that aside

I mean my answer to that is

and I think other people have echoed

this of late

um

I mean it's sort of surprising to me my

answer is that

this begins to privilege

a return to the the humanities as a kind

of a core like the center of of of mass

intellectually for us because when you

look at what we're

really good at and

uh

it's among the last things that can be

plausibly automated

uh and if if we automate it

we may cease to care about it so it's

like

learning to write good code is something

that is going to be it's being automated

now it's it is you know I'm not a

programmer but

um you know I have it on good authority

that already these large language models

are improving code and something like

half the time they're writing better

code than than people

that's all going to become like chess

right it's just it's going to be better

than people ultimately

um

so being a software engineer is

something that you know and being a

radiologist and being like those things

it's easy to see how AI just cancels

those professions or at least makes one

person

you know so effective at using AI tools

that you know one person can do the work

of 100 people so you've got 99 people

who don't have to be doing that job

um

but

creating art and you know writing novels

and being a philosopher and uh

talking about

what it means and to live a good life

and how to do it it's like that's that's

something that if we

we have we have to look at those we have

to look at

where we're going to care that we're

actually in relationship to and in

dialogue with

an another person who's who we know to

be conscious

right like where we don't care about

that we're not going to care we're going

to want just the best version of it like

I don't care if the cure for cancer

comes from an incentive and AI

I do not give a I just want the

cure for cancer right like there's no

added value

that where I find out okay the person

who gave me this cure really felt good

about it and he's you know he had tears

in his eyes when he figured out the Cure

every engineering problem is like that

we want safer planes we want you know we

just want things to work we're not

sentimental about the the Artistry that

went into all of that

uh and when the difference when the gulf

between the best and the mediocre gets

big and consequential

we're just going to want the best we're

just going to want the best all the way

down the line but what is the best novel

right what is the best podcast

conversation what is it and can you

subtract out

the the conscious person from that and

still think it's the best and and so

like so someone once sent me a um

what purported to be I didn't even

listen to it so I'm not even sure what

it was but it looked like it was an AI

generated conversation between

Alan Watts and Terence McKenna right

both guys who I love I remember I didn't

know either of them but fans of both

have listened to hundreds of hours of

both talk as far as I know they never

met each other it would have been a

fascinating conversation

um I realized my when I looked at this

YouTube video I realized I simply don't

care how good this is because I only

care if it was actually Alan Watts and

Terence McKenna talking like a

simulacrum of Alan Watson and Terence

McKenna in this context

I don't care about right so another use

case I I stumbled upon

I was playing with with chat GPT and

I asked it you know the causes of World

War II you know give me 500 words on the

cost of World War II because it gives

you this perfect little you know bullet

pointed essay on the cause of World War

II

that's exactly what I want from it

that's fine that's like I don't care

that it was there was no person behind

that typing

but when I when I think well

do I want to re read

Churchill's you know history of World

War II it's on my shelf to read as I you

know it's like I'm one of these

aspirational sets of books haven't read

it yet

um

I actually want to read it because

Churchill wrote it right like that

that's why and if you could give me an

AI version of Churchill saying this is

in the style of Churchill it's very even

Churchill Scholars say this sounds like

Churchill

I actually don't care about it like like

that's not the used I I'll take the

generic use of you know give me the cost

of World War II

the fake Churchill is profoundly

uninteresting to me the real real

Churchill even though he's dead is is

interesting to me so the rebuttal I give

here and this is what my mind is doing

is yeah saying this

the distinction you're you're presenting

the the difference I see is that in the

case of the conversation between two

people you respect that has been

generated by AI someone has signaled you

that that it is fake

if you remove that because say Churchill

thought yeah why would I write a book

when I could just click a button and

this thing will write it in my in my

voice in my tone of voice with my you

know with the entire the entire back

catalog of things I've written before

and it will produce my my account and it

will save me time so I'll just click a

button my publisher maybe will do it for

me and then I'll sell that to Sam on the

basis that it is

um my thoughts which I imagine I I can

imagine a very near future if we just do

it by percentage how many books are

going to be increasingly written by

artificial intelligence to the point

that when you look at a shelf I imagine

at some point in the future if the

intelligence does increase

um by any measure that most of it would

be words strung together by artificial

intelligence and it will be selling

potentially better than the words

written by humans so again when we go

back to the conversation with your your

children

there might not be a career there either

because artificial intelligence is

faster can produce more contest and

iterate on whether it sells better

clicks gets more clicks it can write the

headline create the picture write the

content and then I can just take the

chat because I put my name to it

yeah

so I go even in that regard what remains

well so in the limit

what I think we're imagining is a world

where

and so none of the terror none of the

terrifyingly bad things have happened so

it's just all working we're just

producing a ton of great stuff that is

better than the human stuff and people

are losing their job so we gotta we got

a labor disruption but we're not talking

about any other kind of political

catastrophe or or you know cyber

apocalypse

um

much less AGI destroying everything

um

then I think we just need a different

economic assumption and ethical

intuition around the value of work I

mean our default Norm now

in a capitalist Society is you have to

figure out something to do with most of

your time

that other people are willing to pay you

for

right you have to figure out how to add

value to other people's lives such that

you reliably

get paid otherwise

you might die right like we've got a

social safety net but it's it's pretty

meager you know we're not we're there

are cracks you can follow through you

could wind up homeless and we're not

going to figure out what to do about

that we're all too well you know and

um

your so your claim Upon Your Existence

Among Us

you finding something to do with your

time that other people will pay you for

right

and now we've got artificial

intelligence removing some of those

opportunities creating others but in the

limit

and I do think it is different from I

think analogies to other moments in in

technological history are fundamentally

flawed I think this is a a technology

which in the limit will replace jobs and

not create better new jobs in in their

wake right it's just this just cancels

the need for

for human labor ultimately

and it's strangely it replaces some of

the highest status most cognitively

intensive Jobs first right you know it

replaces replaces Elon Musk before it

replaces your electrician or your

plumber or your masseuse way before

right so we have to internalize the the

reality of that if again this is in

success this is not it's all good things

happening right

um

and we have to have a new ethic we have

to have a new economics based on that

ethic which is

you know Ubi is one solution to this

like you shouldn't have to work to

survive right Universal basic income

yeah there's there's so much abundance

now being created

we have to figure out how to spread this

wealth around right we've got a cure for

cancer over here we've got perfect

you know

photovoltaic uh

driven economies over here where it's

like we've solved the climate change

issue you know we're just pulling wealth

out of The Ether essentially

um we've got you know nanotechnology

that is just birthing whole new

Industries yeah but it's all being

driven by AI we don't you know there's

no room in this whenever you put a

person

in the in the