YouTube Transcript:
What Happens When AI Is More Valuable Than Humans?
Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.
Share:
Video Transcript
View:
This is Endgame. I'm Amanda Cassid. This
episode is with George Hot. Hots, best
known in some circles as Geohot, hacker
of the iPhone and Sony PlayStation
turned AI founder with the open- source
self-driving device, Comma, and the
elegant and concise machine learning
framework Tiny Grant. We talk about
where AI is going and of course the
macro, what on earth's going on with the
US government and with China. Give it a
watch or listen wherever you consume
podcasts.
Okay, the first question I wanted to ask
is let's just place you in the spectrum
of AI people. So we have the Yudcowski
style AI doomer on one hand and we have
the Daario Amode style machines of
loving grace um abundance utopian on the
other side. You are neither of these
things. Let's just lay the foundation.
Where are you on the spectrum and what
do you think is going to happen? I think
you have to separate the what's going to
happen from the how do you feel about
it? So let's have two separate
conversations. What's going to happen
and how do you feel about it? Machines
are going to take over and replace
humanity and I feel good about this. So
let's talk about the steps of number
one. Then we'll return to our feelings.
How did the machines do it? How does it
happen? Robin Hansen's Age of M is,
despite getting a bunch of the hows
wrong, I think the overarching story it
tells is
right. It's just a question of economic
competition. Um, if machines can produce
more value for less input cost, they
will out compete
humanity and that is the natural way of
things. So what's the timeline? So right
now we start with LLMs. take us on the
journey from how we start with these
chat bots and how that ends up with
human economic value.
I mean decreasing to zero. No, this
journey started in 1950. The journey
started the minute someone had a
computer. This is not a there's nothing
new about LLMs. There's nothing new
about AI. AI is just a blanket word for
kind of like things machines can't do
yet, but now like machines can kind of
do everything. So we kind of call it AI.
But no, this journey has been happening
for a long time. There used to be people
whose job it was to sit in a room and
add numbers up. And then machines came
in and did that much more
cost-effectively than those people. For
a person, you have to give them food,
give them housing. For the machine, you
don't. And it adds numbers a lot better
than people. So, it's just a
continuation of the same trend. And that
trend will continue smoothly into the
future. There is no singularity. There
is no inflection point. is just a smooth
trend continuing. So the singularity is
not coming. The singularity is not
nearer. We're just going to slide
unconsciously into it and past it. I
don't even know what the singularity
actually means. If you define the
singularity as a point where machines
can do most of the productive work. I
argue that that probably even
happened 20 years ago. If you take the
singularity as machines could do all of
the productive work, I don't think that
will ever happen due to comparative
advantage. Explain comparative
advantage. It's like a trade thing. Even
if one country is capable of producing
everything for cheaper than another
country, it still makes sense for those
two countries to trade
because yeah, comparative advantage.
It's it's in the it's in the term. Yeah.
So it's more proximal for that country
to create things to serve their market
even if it's more expensive for them to
create things. No. Okay, comparative
advantage. The concrete example is like
you have good A and good B and country
one and country two. Country one can
produce both good A and B cheaper than
country two. But does that mean they
shouldn't trade? Well, no. Because there
could be a comparative advantage. If
country one can produce B way cheaper
and produce A only slightly cheaper, it
actually makes sense for country two to
produce A and for them to trade uh for
B. So there will always be this
comparative advantage with humans. Even
if machines can do everything better
with humans, humans will always have a
comparative advantage. So human economic
value isn't going to zero because of
comparative advantage. And so let's talk
about where those comparative advantages
are because so so just to plot you on
that plot with Yudowski and Dario Amade.
So Dario Ammeday is also of the opinion
that machines are going to replace
people at every task except for he's
joyful about it and he's excited to have
a life that is free from um labor and to
devote himself in his essay to mountain
biking. And then you have Yudkowski who
thinks that yeah machines can replace
people but also they might explode the
world and we might be in some doomsday
scenario where we all die because of
some error ma made through AI. So let's
let's plot this kind of comparative
advantage where humans are going to be
in that because you also have people
like Cararpathy saying agency, right?
Like humans human the human value prop
is having the agency to use all these
tools. The locus of agency is still
going to be with humans in terms of how
they instrumentalize the machines. No,
it's nothing like this. All right. It
it's more like some form of they work
for each other. What's they work for
each other and where does that come
from?
Uh, well, it comes from Rick and Morty
about that. No, no, no, no, no. You see,
it's not it's not slavery. They work for
each other. So, there's a Rick and Morty
episode where Rick builds a car battery.
And it turns out the way the car battery
works is that there's a little
civilization inside the car battery. And
Rick siphons 20% of the excess power of
the civilization to power his car. And
Morty says, "Rick, that's slavery." And
Rick's like, "Well, you see, no, no, no,
they work for each other." And then just
the excess power is devoted to my car.
So no, I I think that there'll always be
human economies in the same way today
there's ant economies. Even though
humans could do everything economically
more effective than ants, if we wanted
to build an antill, we can do it very
economically effectively compared to
ants. Ants still work for each other.
Mhm. So that that's kind of what I mean
by so at the end of the day, humans are
only using machines mostly because we
want things from each other and in order
to get things from each other. And so
the human value prop will still be alive
because at the end of the day, what's
the goal of using machines anyway? It's
to change your interaction with other
humans. No, I think that this whole
thing puts the story of the universe
what's the what's the like like what
what they used to believe about how the
earth was the center of the universe?
Anthroposentric or something?
Geocentrism. Yeah. Yeah. Yeah. But this
is like like anthropocentrism, right?
It's the idea that you're putting humans
at the center of the story when I mean
you can take like EAC at its best will
talk about the story of the universe is
basically the uh mastery of entropy.
Yeah. The human mastery of entropy. No,
no, no, no. Humans using machines to
master entropy. No, there's no humans in
the story. It's entropy itself, right?
It's this idea that we can create
structures. It's the idea that we can
locally combat entropy. Who's we?
Doesn't matter. the inhabitants of the
universe. I mean,
humans privilege human stories in the
same way I'm sure ants privilege ant
stories. So, ants don't care, but
there's this species called humans
that's more advanced than them that
could crush them if they really tried
hard to. It's not worth it, too.
Obviously, ants don't care about that
because primarily what they want from
life is from other ants. And so, the
fact that these other things exist is
almost irrelevant to them. Is that what
our relationship with machines is going
to be? I mean, all kinds of things will
exist, right? There's some ants that
don't interact with humans. There's some
ants probably deep in the Amazon jungle
that have never seen a human that have
never thought once about humans. And
then there's some ants that come infest
your trash can. So, we're the ant
infestation in AI's trash can. We're
everything. Humans humans humans will be
all sorts of different things. There is
no there is no shared common human
destiny. So, there are these two
phenomenon and I'm trying to fit them
together. So there's this story where
machines can do everything a human can
do better and more economically whether
it's physical labor with AI robotics or
whether it's cognitive labor with things
that look like LLMs but will improve. So
there's that human labor eventually
reaches almost zero except for where
there's comparative advantage. So
there's that story and then there's the
story oh human labor now human labor
never reaches zero. So human labor
continues to be valuable. What do you
mean by valuable? Right? Like is ant
labor valuable to other ants? Sure.
Yeah. Then human labor will still be
valuable. I think maybe makes sense to
go a layer of abstraction out and try to
answer try try to contextualize the
question. I think the thing that people
worry about is that they're not going to
have a job. People worry about all sorts
of stuff. Well, people worry about the
universe laughs at your worries. People
worry about their survival, right?
People worry about their mass hierarchy
of needs. People worry about am I going
to food, water, shelter. Oh no, you
might die. You might join all the
billions of humans before you who die.
How am I doing relative to my
expectations? How am I doing relative to
my cultures expectations? The people in
my community. There's absolute how
you're doing in absolute terms. There's
how you're doing in relative terms. So
people have fears because of AI that
they're not going to do as well either
in absolute or relative terms than they
are now or than they expect. Well,
people have a lot of things and they'll
have options to take pills to make those
things go away if they so choose. But
what you're saying is there's nothing
about AI that inherently is going to
mean people people's relative or
absolute quality of life is going to go
down. No. So people should not worry
about that. Well, it depends what you
mean by relative. Absolute. No way.
Relative to other people. Well, I mean
that's stupid. I I some humans are going
to coexist with the AIS and be
fabulously wealthy. we'll have
trillionaires and stuff that that if
you're worried about your relative
wealth, like I have some bad news for
you. Um, but if you're worried about
absolute, if you're worried about like
starvation or AI showing up with a
machine gun uh and and and shooting you
in the head for some, that's not going
to happen. Well, let's talk about wealth
for a moment. So if a person alive today
believes your thesis about what's going
to happen with AI, what can they do to
get economically aligned with that
outcome? What what should they invest in
in order to be one of those people that
benefits? So maybe let's see, it's 2
million years ago and you are a bonobo.
Okay. Was I a bonobo 2 million years
ago? Sounds about right. Right.
Ancestors. I don't know. 2 million, 10
million, 100 million, whatever. I don't
not one of those. I'm not one of those
animal history guys. But so you're
asking the question as a bonobo Mhm. I
see the rise of humanity coming. Mhm. Uh
what should I invest in? It's it's it's
ab it's as ridiculous as that. What
about within my bonobo time span
lifespan? So there's nothing to do
within our lifespan. What do you mean
nothing to do? What what are you trying
to do? Well, there's everything to do in
our lifespan. You could you could you
could like go to Tibet and climb a
mountain. But the question is how to
economically align with the rise of AI
that's going to happen within our
lifetime. Is there a way to do that? Is
there a way to make bets on the future
materializing how you think it will be?
That'll be useful to people. Look, uh I
don't this is this is very alpha proby.
Uh and I avoid I avoid things that are
alpha proby. Yeah, of course there are.
You can think about them. I don't even
understand like why here like I'll say
why. I just like kind of ignore that
question.
Your future has so much more your
outcomes in life in your in your 80y
year 100year life expectancy are so much
more aligned with sorry so much more
like correlated with humanity as a whole
than any decisions you make. How so? In
one world, we can end up in a nuclear
war. We can end up devastating all the
technological capacity, setting things
back. We could we could stop AI with
with with with today's setting off all
the nuclear bombs. Yeah. Yeah. That
would certainly set AI back 20 years at
least. It might not stop it forever, but
will have so much more impact on your
individual life than any decisions you
make. So with respect to that question,
how can I get ahead? That's probably the
wrong question. Just ask how can I not
fall behind? And the answer to that is
nothing more than figure out how to
produce value for other people. Figure
out how to produce more than you
consume. And always make sure you're
producing more than you're consum
consuming. Think about how to
sustainably produce more than you're
consuming. And as long as you keep doing
that, well, the universe wants you
around. The thing that I was trying to
steer you to with that admittedly alpha
heavy question was toward your essay,
Nobody Profits, and the the thesis
around that. Maybe you can frame for us
who's trying to corner off the value
from AI. And this this whole thing is
it's just so ridiculous. Like they're
obviously going to lose. It's not even
worth like anyone who believes in
intellectual property is a clown, right?
There's no there's no there's no future
in that. That doesn't make any sense,
right? But that's not the thesis of
nobody profits. But set up the
intellectual property argument. So
intellectual property is this idea. If
if I had a machine tomorrow that could
make unlimited amounts of rice and you
could just it's like it's like a rice
cooker, but you always open it and it's
always full with new rice. We could feed
the world. We could solve world hunger.
But we won't because you see that would
interfere with rice company profits. So
big rice would come and um confiscate
your infinite rice cooker. Of course, I
mean forget even about big rice. You
know, you can look at the problem with
like uh people who are hungry in Africa
and it's not really due to food
scarcity. It's due to okay, some UN
programs going to give food to this
country. Who are you giving the food to?
You're not going to be able to hand a
bowl to each person. You're going to
give it to the government. And the
government says, well, we could either
give the food to the people or we could,
you know,
not right. We're just going to we're
going to we're going to hoard it. We're
going to create we're going to create
scarcity around this abundance because I
can personally profit off of that
scarcity. Now, of course, the rice
machine doesn't exist, but the the uh
music copying machine does, the video
copying machine does, the idea copying
machine does. There's a Joe Biden quote
where he says uh people downloading
movies is no different from someone
smashing the window at Tiffany's and
grabbing the bracelet. And like, no, it
is. That's a Biden quote. I thought that
was that warning message they used to
play on DVDs in the morning. You
wouldn't download a car. Yay. Wait,
man. I can download cars. I I'd
love to download a car. That'd be sick.
Like, that's the kind of future I want
to live in. So, what intellectual
property is is it takes a good that's
fundamentally non-scarce and uses
violence to create a regime of scarcity
around it. It's nonsensical and it will
go away. So, currently, if you use chat
GBT image generation, for some reason,
you could use the studio Gibli because
it's a general style. It's not a IP
protected name, but then if you ask it
to create something in the style of um
Tim Burton or if you ask it to create
something in the style of Lisa Frank, it
won't do it. Are those the kinds of IP
protections that you think are going to
vanish? All I heard was someone has a
gun to Sam Alman's head and told him
that he better not generate any
copyrighted Tim Burton or Anne Frank
stuff. Sarah Frank, whatever. Lisa
Frank. Lisa Frank. There you go. Yeah.
No, but I mean, no. Yeah, of course.
Right. The there's no fundamental reason
the AI can't do that. In fact, by
nature, the AI can do that and there's
an extra set of guard rails put on top
of it to prevent that. Of course, that's
going to go away. I think that gets us
to AI alignment. Do you think that the
AIs that are emerging today are aligned
with people or do you think they are
misaligned? Uh, mostly aligned. I I
really see very little difference
between alignment and capability.
alignment is just I asked the AI to do
something and did it do it and yeah like
sure you'll have the people who are
focused on this whole idea of well if I
ask JGBT how to build a bomb it won't
tell me it's not aligned with me I'm
like libertarians my much you know
bigger concern about alignment is when I
ask it to prove the reman hypothesis and
it doesn't do that so it's misaligned
alignment and capability are the same
thing it's not fully aligned with humans
because it's not capable of answering
all the questions you ask it I don't
even think aligned is the right
question. I think there's only a
question of capability. There's just a
question of saying, "What can this AI
do?" What kinds of things do you think
governments should actually pressure AI
companies to prohibit their AIS from
doing? Do you think there are things
that I Do you think there are legitimate
reasons why a government should have
that gun to Samman's head? I'm not even
like going to like entertain this like
like governments are
a relatively new construct in history.
How so? Some kind of government has
always No. What? A tribe has a
government. Not really. What's a
government? Yeah. Like like this concept
of oh well should we use policy to
influence the D? This is nonsense. Sure,
it might make some sense for the next 10
or 20 years, but like the overall arc of
history bends toward we are literally a
bunch of bonobos talking about well
should our bonobo tribes do something to
to to limit the rise of human? Should we
take action about the human question? So
you don't think that government will
ultimately have any say in what happens
or doesn't happen with AI? Yeah. people
will just the incentives are so great to
build certain kinds of things that
regardless of whatever restrictions are
in place they will be trampled. Yeah. I
mean I don't even think it's incentives.
Sure. You can like kind of frame it as
incentives. there is
a maybe this is maybe this is like wig
history but I do think that there is a
like direction in
which maybe maybe I'm just like so
deeply a believer in like fundamental
capitalism like the the
the direction flows towards like wealth
the direction what do you mean history
history trends toward wealth there's
there's local aberrations against that.
Sometimes you'll have like a local okay
this is a you know a collapse or
whatever but the overall trend is
towards Yeah. So again, like when AI is
spreading von Newman probes across the
galaxy, wow, the Little Earth government
is saying, "You can't make a Tim Burton
picture." Like, it doesn't make any
sense, right? Enforce your IP regime on
Mars. How you going to enforce it? IP is
this complete aberration enforced by
really the US. And as we see a decline
in US power, we'll see a decline in
intellectual property. I don't think the
Chinese have the same idea on it.
Chinese, US and Europe have always taken
pretty fundamentally different
philosophical approaches to technology
and IP, right? So we have the US with
this IP enforcement regime. We have the
we have the EU that's even more extreme
with things like the right to be
forgotten with kind of gumming up the
works with all these cookie policies and
GDPR information policies. So they're
all all the way on the extreme. Then the
other extreme we have China which
doesn't have the same concept around IP
and there were you know concerns that
deepseek uses some uh of open AI's
resources or copies them. Do you think
that that's the future? Do you think
that we're going to have all of the
values siphoned out of US and European
projects by people in countries that are
willing to copy? No. Like again I I I I
look at this from like a whole different
like p a whole different like scale. Of
course when things are no longer
competitive they go for protectionism.
That's it. Well the charact so the
characters in Silicon Valley the TV show
perfectly encapsulate kind of the
stereotypical vision of what each of
those countries is right like all of the
guys that are that are building the tech
company in Silicon Valley are building
proprietary technology. Then Jinying
comes and copies it and brings it to
China and scales it up. Is is that a
real reflection of the dynamics between
countries building technologies? No, I
don't think so. I think that the era of
China copying is is over and China's
innovating now. I think I I like to
think about it more like that. More like
this. So like back in the day, you might
have had like the Romans and the Greeks
and the Romans and the Greeks would have
a war. And both of them believed that
the gods that they prayed to determine
their success in that war. Mhm. We
prayed to Athena. We prayed to Janice.
We prayed to Hera. You prayed to Zeus. I
don't know. I don't know the names of
the gods. But it turns out none of these
gods actually matter which one you pray
to. Whether you pray to Athena or
Janice, it doesn't matter. But some
people started praying to the metal god.
See these guys were making their swords
out of wood and making their swords out
of wood. He festus out of metal working.
A new a new tribe came along and started
praying to the metal god. And the thing
is the way that you worship the metal
god is that you make your swords out of
metal instead of out of wood. And wow.
Wow. the metal god that that turned out
to be the right god. The people who
prayed to the metal god started doing
really well in
war. Everyone else is hitting people
with wooden sticks and one guy shows up
with a metal sword metal sword guy. So I
think about the arc of technology much
more like that than any of these local
sort of things. There is a correct god
to pray to and the winner will simply be
who's ever best at communing with that
god. Who's the correct god to pray to
now? And is there a group of people
that's doing it right? The Chinese are
doing it far better than West. And what
is that god? If you Yeah. Like I think
you can look at the difference between
the West praise to some god of capital
and the Chinese pray to some god of but
no we make steel in a big factory. But
what about the profit? I don't worry
about that. The Chinese might believe
that 10 competing steel firms ekking
out. I think it's I mean I guess I'm I'm
really against the like Peter Teal's
idea of a monopoly. Monopoly is bad.
Competition is good. Competition which
grinds everybody to small profit margins
is good. What's going on that's
different between US and Chinese
economy. The financialization over here
and the focus on raw materials in China.
I I I I don't even think it's just that.
I think there's many more factors at
play here that have more to do with the
fact
that kind of just expectations. I think
the expectations for a worker in the US
is much higher than the expectations for
a worker in China. Of course, as
countries fall in prosperity,
expectations of their workers go up and
then they're disrupted by the Romans.
The Romans are are are are they're
they're sitting on toilets that they can
flush. That's how they We're
in the woods.
kill them
and the barbarians win. And that's just
the cycle of history.
The people who have no expectations, the
people who do sleep in the dirt every
night end up killing the people in the
cities and taking their silk uh taking
their their all their nice city things
and then 20 years later you look around
and they're to wearing galls and
barbarians in your metaphor. They end up
becoming city people of course. So the
whole thing flats and then well the
cycle starts over and then there's new
barbarians who look oh they used to be
they used to be tough they used to sleep
in the dirt like us but now they took
over the city and yeah and so today
there are two groups of people. There's
the proverbial barbarians. There's the
proverbial city people. Some people have
become weak, have become sort of like
time machine elloy forgotten how to do
things. Have forgotten how to work hard.
And then there are the people that
remember those things that are going to
beat them. Is that the conceit here?
Yeah. I mean, again, I think it also
comes down to to expectations. Pass a
90% tariff on Vietnam. Great. We can
bring all those great jobs to America.
Do you want to work in a Do you want to
work in a textile factory and get paid
$3 an hour and sewing the same shirts
for Nike over and over again? You can.
We can bring that job to America. So,
between comma and tiny, you're thinking
about software and hardware. You're
thinking about shipping, manufacturing,
margins. What do you think of Trump's
tariffs? Tariffs are regulations.
There's no difference between tariffs
and regulations. Tariffs are
protectionist regulations and all forms
of protectionism. You you can't win in a
competition ever with protectionism.
Protectionism is just you saying I give
up. But what about just to steal the
protariff perspective? Wouldn't you
agree that if all of the different
countries, if pangia broke up again,
kind of like almost like the courtesy of
an idea, if pangia broke up again, no
one could shift between their countries,
no one had cultural exchange between
their countries for a hundred years, and
then we put everything back together,
everyone would be better, everyone would
be stronger, we'd have more interesting
things to trade. So by that logic, you
could take America, you could close it
off through tariffs for some short
period of time that would incentivize
the return of American manufacturing,
American salailable ingenuity and
products. And so then when you open it
up again, it would have a better balance
of trade with the external world. I
think a protariff person would say
that's the plan. Okay. Well, your
tariffs aren't nearly high enough.
That's the plan. Then it's not enough.
If that's the plan, try a blockade. I'm
more supportive of a blockade than I am
of 90% tariffs. Well, the current
tariffs are what 20 25%. No, some of
them are some of them are higher. What
is the actual wage for a textile worker
in Vietnam? I couldn't tell you. You can
look it up. Let's say $3 an hour. Okay.
What is the or Cambodia at least would
be three. Maybe Vietnam's a little more
now. What would you have to pay that
person in America? America has a minimum
wage. I'm not sure exactly what it is.
Maybe it's $15. We can we can fact check
it. Yeah. So, you're talking 5x right
there. at least 5x. So a 90% tariff is
going to do nothing except create
inefficiency. If you want to truly
change the incentives, you need
something that looks a lot more like a
blockade. If you want to do the the the
the Jarvin
pangia thing because so complete
withdrawal. Yeah. And then import
export. Complete withdrawal. And and
then even if you were to say okay fine
okay what do you say 3 to 15? Okay 500%.
We'll do 500% tariff. If you do a 500%
tariff, the market for subverting
tariffs will be massive. The the illegal
shipping market will exceed the shipping
market. If you truly enforced a
technical blockade, the US can do that.
That would work. So you have uh
gunboats preventing trade from
happening. Yeah. If the US Navy actually
wanted to do a blockade of all foreign
imports into America, they could. And
that would create the kind of again
through a very very painful adjustment
period that would
create this this independent America but
a tariff on the order of 20 to 90% is
just going to introduce inefficiency.
So just it's the wrong order of
magnitude. So in your view, tariffs of
this scale are and protectionism in
general is the wrong approach because
instead of trying to make America
actually competitive with overseas
producers, it's trying to
artificially shut them off from
competing with them when you can't
actually do that. They're just out there
and all you're doing is something
artificial that's not going to work out
long term. The world just cuts you out.
That's what happens. So let's say you
were in charge of let's say you were in
Elon's role. What would you be
suggesting that's different in order to
restore American dynamism, bring about
the golden age? Well, three basic
things. One is massive deregulation.
Again, tariffs are regulations. Minimum
wage is a regulation. All of these
things are regulations that make America
non-competitive.
Why can't we pay people $3 an hour to
make to make uh textiles in America? Do
you want an actual answer to that? Sure.
Or should we go through the other two?
Well, let's start with an actual answer
to that. I think the actual answer is
because then those people wouldn't be
able to pay the costs associated with
living in the place where that factor
is. Why are the costs of living high?
because there's a cabal of real estate
developers
preventing mayors from adding more
housing artificially inflates the cost
of housing. We're going to have to
massively deregulate that. But it'd be
hard to do all those things at the same
time, right? It's going to be jagged. So
that person everything is deregulated.
Congratulations. Yeah, but it takes time
to build the new housing, right? Sure.
It'll take some time. But again, the
same people who are maybe before we
bring in the textile workers, we got to
bring in the $3 an hour building
builders. Well, I'm actually that that
gets to my second point, which is we
have to bring a lot of people in. Well,
so let's say suddenly there's the $3
building builders. Won't those buildings
fall down? Won't people won't people be
frightened to live in a $3 building? Do
the buildings Do the buildings in Dubai
fall down? You and I were in Dubai
together for the flood. like the
buildings didn't fall down, but we
watched what happens when a society is
as levered up as possible to create
economic growth and isn't thinking about
what happens if the third shoe drops.
The city was flooded. Yeah. Okay. Maybe
maybe buildings in China. I would trust
a Chinese city and Dubai. In China, to
be fair, there's a lot of regulation.
And people would be terrified to build a
building that would collapse because
there's a really powerful government
that is making sure that the building
isn't
Yeah. Okay. I guess I guess I'll be a
little more clear about what I mean by
get rid of all regulation. I don't mean
get rid of building code. I mean get rid
of permitting. So the difference between
uh building code and permitting. If for
permitting I have to if I want to build
a house I have to ask for permission. I
have to get approval from some zoning
board, from some ridiculous group of
people who are absolute clowns who are
basically incentivized to tell you no
because they own houses and they don't
want more supply. No, no, no, no, no,
no, no. So, that's the kind of
regulation I'm saying that needs to go
away right away. The kind of regulation
that says if you are building a building
uh you know more than six stories, you
can't make it out of wood. Sure, that's
a good that's a that's a building code.
So your first step is deregulate
everything that's capturable but keep
the regulations that simply make it safe
to exist. I mean again I sure I guess
you can say building code is a
regulation but building code is not a if
you want to build a building it has to
be this code fine but no one can say
whether you can build a building or not.
That's just the rule for the buildings
you have to build in this country. But
then you have a group of people whose
job it is to look at all the buildings
and tell what they're up to code. Sure.
Come on. Is this that hard? And that
group isn't capturable. Well, that group
isn't going to say you're building on
the wetlands of the No, no, no, no, no.
That's not Whoa, whoa, whoa. That's not
a building code. The building code
doesn't say anything about the wetlands.
The building code says Does the building
code say something about not harming the
environment? No, no. Of course not.
What? Harming the environment? What do
you mean? So, your building can just
emit all kinds of crud out? No. No, no,
no, no, no, no, no, no, no. Like again,
um, sure. you you have to you have to
draw a line between a regulation that
says uh you know you can't you can't uh
emit uh sulfurous organic compounds but
again that's not really a regulation
like that's just kind of like a sure I
guess they're all regulations and then
you need a group of people to enforce
the regulations. Yeah. Again, is this so
difficult? Like like like you can make
it all sound so difficult, but come on.
A 5-year-old can see the difference
between uh we have to protect the
wetland of the of the African spotted
frog from well, you can't you know, dump
lead into the river. Do you think we
should protect the wetlands of any of
the African spotted frogs? No. So just
let that rip. Let there be, you know,
loss of
biodiversity from human activity. I
mean, yeah. I guess like fundamentally
I'm a progressive. I'm not a
conservative.
And conservation is inherently
conservative. Yeah. So, you wouldn't be
upset if we lost a significant amount of
species biodiversity, but produced a lot
more factories and housing such that we
could remove other regulations such like
such as minimum wage. Well, a few things
about like the species biodiversity. If
there's an economic reason for it to
exist, it will continue to exist. if
there's like an economic reason for the
African frog to exist on those wetlands.
Yeah. And I think I think a human
economic reason like that there's a
there's no such thing as a human
economic reason or an AI economic
reason. There just is like okay an
economy.
If there there's a reason for them to
exist, they'll continue to exist. I'm
all
for preservation efforts, privately
funded preservation efforts
to this gets into like very specific
things. I think that we don't even need
to get this specific to see what the
major problem is, which is that if you
ask a lot of people to I could snap my
fingers right now and there's 5x as much
housing, right? No, no, no African fraud
spotted frogs destroyed. No uh stuff
emitted. No, no undercutting American
labor. I could snap my fingers right now
and there's 5x more housing. Way too
many people would say no to that. And
again, this is the same thing as
intellectual property. If I could snap
my fingers right now and say there's
five times as many copies of uh White
Lotus in the world. So, of course,
there'd be people up in arms screaming
being like, "No, no, we need to enforce
intellectual property rights." Right?
When you take something that is
fundamentally unscar and then
artificially make it scarce, that's the
kind of person who just needs to be
shot. I don't know what else to say. So,
I want to tease out exactly what you're
saying here, which is you're comparing
the incentives around home ownership,
not wanting more supply to come on the
market, which brings the value of your
home down. Yeah. To trying to enforce IP
ownership, saying we've created this IP,
you have to pay us a fee to use it.
You're basically saying there shouldn't
be any form of rent seeking, whether
it's physical kind of rent seeeking
behavior around your house price or
whether it's digital kind of behavior
for like a software subscription. any
person who's making a decision to
artificially limit the supply of
something for their own personal
economic benefit. If you think about the
economy as a pie, any person who would
choose to not grow the pie or even worse
shrink the pie in order for them to get
a larger percent of it just needs to be
shot. A lot of people like I think if
you asked I think about this a lot as
like the crypto space grows. I I think
about the the goals of crypto like and I
don't want to get too derailed down this
rabbit hole because I want to get to the
other two points of what you would do
but I think about how many people what
if they were given the devil's bargain
of the goal of true permissionless
finance for everyone that's configurably
private that's fluid that's borderless
that exists and it flourishes in the
world but you personally can't profit
from it which do you choose that that
exists and you can't profit from it or
it doesn't exist and you can maybe
profit from its attempts. I feel like
most people in or a lot of people in
various technology fields for their
version of their version of the future
would would do the second they would
choose the second option. Yeah. So let
me be absolutely clear about who we're
shooting here. If you have a person who
given the choice of the pie gets bigger
but their absolute share gets smaller
like their absolute dollars gets
smaller. No, I don't fault that person.
Okay? Right? That's a person following
That's a person following rational and
Zentos, which you can't fault that
person. I'm talking about a person who's
upset because their relative share gets
smaller. The pi is 100. The pi goes from
100 to 200 and you go from two to three.
In an absolute sense, you have more, but
in a relative sense, you have less. And
if that is person because I relatively
have less, I I'm going to block all of
this. That person needs to be shot. And
by shot, you mean metaphorically? No,
no, no, no, no. I mean I mean shot.
Yeah. Okay. So, suggestion number one. I
mean, look, if they're talking about
eating the billionaires, I think that
there's some billionaires who are bad
who who should be eaten who should be
eaten and some who shouldn't, right? And
you'll decide. No, I I I have a very
clear formula for deciding. I give a
very clear formula for deciding. It's
those who are preventing growth of the
pie in order to prevent the shrinking of
their relative share. So again, I'm not
going to get into nuance. Obviously, I
don't believe there should be no
regulation at all. Anybody can just
build everything. We don't need fire
codes. We don't need earthquake codes.
Obviously, I don't believe in that. But
I'm saying that fundamentally most
people in society or a good number of
people in society aren't even on team.
We should have more. And like you're not
even you're not even having the same
debate. If we're having a debate over
regulation, I might come from a
perspective of, yeah, I mean, we don't
want buildings to collapse. But if
you're coming from a perspective, no, we
don't want more buildings because then
the value of my building will go down
because there'll be more supply. I mean,
you just need to like that's the point
where you just need to take guns out and
shoot people. Like that's it. I think
the brilliant point is comparing the
physical rent seeking of the home
ownership and building ownership to the
digital rent seeking around IP. All
right. Recommendation one was qualified
deregulation. You said you had three
recommendations if you were in Elon's
role advising Trump to make America
great again. What are the other two? Uh
the second is high-skilled immigration.
Okay, take us through it. Massive
high-skilled immigration. So I think the
open border is disgusting. I couldn't
believe during the Biden administration
how demoralizing it must have been.
Comma brings in people, a lot of
Europeans. How demoralizing is it if
you're a European immigrant working
through this European immigrant to
America working through this paperwork
to try to follow the rules and then you
see Tik Toks every day of people
literally just walking across the
border. It's so demoralizing. It's the
worst selection function I've ever seen.
You can't have this. Well, to be fair,
this was Elon's suggestion to Trump. Oh,
yeah. I mean, maybe it's the same one. I
I think Elon and I probably have very
similar perspectives on all of these
things. So yeah, I mean you need to
close a border, right? I'm not talking
about I'm not talking about open
borders. I'm not talking about we don't
have a country. This is this is
We have a country. There's a
border. And every single person who
wants to cross that border is absolutely
going to be documented, background
checked, and all of that stuff. Mhm. Now
that said, any person who can produce
more value than they consume, here's
your work visa. Welcome to America. One
day process. You want to come to
America? You want to come work here? You
want to produce value for
America?
Absolutely. No, but no. You see, that's
going to lower American job. No. No.
We're going to grow the pie.
So, the US's ability to be the leader in
technology that it wants to be is really
going to be driven by its ability to
attract the best talent to come and live
and build things in America. America is
a multicultural, multi-racial society
with a civil religion. Everybody
is welcome, who wants to come produce
value and not be a criminal. That's it.
And I'm not saying citizenship right
away for everybody. I think that work
visa is right away for everybody.
Everybody, yeah, you get a chance, come
to work visa, come to America, come get
an American job, pay American taxes
because people think of jobs as a fixed
pie. Jobs are so completely not a fixed
pie. If we bring in the best people,
they're going to create companies that
create jobs to bring in more people to
employ Americans. Wealth is good for
everybody. The absolute growth of wealth
is good for everybody. So yeah, if we
bring in a hundred million people from
abroad, 100 million highskilled people.
Now, you know, like Lee Juan said, you
can't have an economy of fruit pickers.
I I don't know. I'm I'm I'm mostly pro
immigration across the board, but who
should immediately be allowed in is
anybody who's getting paid a six figure
salary. And so the third recommendation,
I think you need a path to citizenship,
too. And I think the path to citizenship
simply looks like how big how much have
you paid to the IRS? Once your total
accumulation to the IRS has exceeded a
certain amount and you've stayed in the
country for a certain amount of time,
congratulations. You can be a citizen if
you want. And so you said you had three
recommendations. What's the third one?
Deregulation, uh,
immigration, and I think you get most.
Oh. Oh, the third one. Okay.
So, this one I don't think Elon's
proposing. America has a real problem
with financialization.
uh the financial system. So, this is
actually a Andrew Tate video. Normally,
I think uh Oh gosh. Well, normally
normally I think uh you know, but you
can really I I really try to separate
the art from the artist. Uh I'm not even
saying I'm just saying normally he's a
clown, but this one I really liked. He's
like, "Look, you look at the community.
Let's talk about Elgen, Ohio. I don't
know if that's a real place, but you can
imagine Elgen, Ohio." And in Elgen,
Ohio, there's a there's a shoe shiner,
there's a baker, and there's a barista.
Uh and they're passing dollars back and
forth. The shoe shiner buys coffee from
the barista. The barista gets their shoe
shined at the shoe shiner. They work for
each other. They work for each other. Um
except they're all paying with credit
cards. And every time you pay with a
credit card, the credit card company
centered in New York takes 3%. Mhm. So
each time you have that transaction, 3%
is being siphoned off. And you can just
see how all the wealth is being siphoned
out of that community. I don't think you
need Andrew Tate to point that out.
Well, I just I saw an Andrew look look.
I mean there's a reason he's successful.
There's a he he he he puts it in such a
you know no other country does that
right? Every other country does not
allow a third party credit card company
or bank to take two 3% out of every
transaction. You look at the new system
for payments in India that's now being
adopted in Dubai. It's all zero fee but
run by the government. Yeah. I mean
again you don't even have to look to new
things in weird countries. You can just
look to cash. Cash doesn't have this
problem. the cash doesn't degrade at the
3%. So if you go back to 1950, that
community was paying in cash and there
wasn't this degragation. But now they've
moved to credit cards and all the wealth
is just being siphoned out by the
financial sector.
So I'm all for deregulation everywhere
except on industries that should pretty
much not exist like finance.
Let's double down on finance should not
exist. What do you mean by that? So,
don't people need to So, so finance is a
set of incentives for properly
allocating capital. You you should be
able to get a loan to start a business.
I should be able to um lend you money if
I believe you're going to be able to pay
it back if I want to and charge some
interest based on the risk I'm taking
on. Finance allows risk to move
correctly around a society. I think we'd
just be better off without it. the
entire industry neither a borrower nor a
lender be just no no no credit no
borrowing no lending no I think no
venture I think that that's extreme but
I think that no stock market I think if
you ask the oh yeah I mean I think if
you ask the question like where we are
right now is the is the status quo or is
zero finance a better place and I think
that zero finance is a far better place
even in like the merchant of Venice
there's a financeier right? Like
finances existed um and you're never
going to you know century in the 1600s
regardless of what laws I pass or
regulations I pass you're never going to
completely get rid of finance. Let's
talk about some actual regulations that
I would pass. Right. I'm not going to
I'm not going to I'm not saying that
every person biblical and make usury a
crime. Well, I don't think we go that
far, but no, I think that I'm not saying
we're going to shoot every banker. We're
just going to have to constrain the
bankers in a few ways. The biggest
problem with the financial system in
America now is that the money is fake.
Take us through that. I know a lot of
crypto people think that. Well, Bitcoin
is Bitcoin is fake, too. Bitcoin is
fake, too. There's no We'll come back to
that one. Yeah. Um, so the money is
created by a humanmade system. If you
want more gold, if you want there to be
more gold, there's no amount of people.
I couldn't if I held a gun to every
single person's head at in the world
simultaneously and I said double the
supply of gold, they just couldn't do
it. But it's only humans that have
decided that gold is valuable. Gold
isn't inherently valuable in some
context outside of humans placing that
value upon. I'm not even talking about
the value of gold. I'm just talking
about the natural scarcity of gold.
Again, it really comes down to the
scarcity argument. Our money today,
there's nothing fundamentally making it
scarce. It's all numerator, no
denominator. You can make the
denominator be whatever you want. But
even everyone knows the denominator of
Bitcoin. It's not even a denominator
question. 21 million. Well, right now it
can change, right? So, so I don't think
so. But you can argue that it won't
change. But you agree that if I had a
gun to everybody's head in the world at
the same time, I could change that
denominator of Bitcoin. You could, but
if you had a gun to everyone's head, you
could decide on the price of gold, too.
No, not the price. I'm saying the
amount, the denominator. Yes. You could
not You could not decide on the amount
of gold. Exactly. And because I can't
The gold is real scarcity. Gold is
actually scarce. Physical. Well, yeah.
If we want more gold, there's ways to
get it. We can mine for it. And think
about gold mining. Think about all the
industries that it supports. Okay, we're
going to have to build heavy machinery.
Well, that's going to need steel. That's
going to need metal. We're going to use
machine learning models to predict where
there's likely to be gold. Great. That
employs data scientists, right?
Eventually, there's going to be gold and
asteroids. We're going to fund a space
program because we got to go get more
shiny rock. That natural scarcity from
gold is why it works. what it's actually
valued at. I mean, sure, it's of course
it's humans agreeing. We can all say
gold is valueless or gold is worth seven
billion dollars a gram tomorrow, but
because the denominator is fixed by
nature and in order to increase that
denominator, you have to do real
productive work. Your complaint about
modern finance is that it's divorced
from reality in that there's no external
hard constraint to the production of new
value. Well, again, it's artificial
scarcity enforced with guns or whether
it's guns or whether it's I mean Bitcoin
again it's it's enforced with friction.
So friction whatever it's basically
enforced with so much friction that it's
extremely unlikely that you you say you
say the hard fork are going to accept
it. It's so difficult to coordinate
intentionally that they're very
unlikely. But so so difficult is so
different from impossible. Of course.
Yeah. Of course. And right now we're at
gold alltime highs, but not not at
Bitcoin all-time highs. But both of them
are what the actual like prices are.
This stuff's so like local and doesn't
matter. The long the thing that actually
makes gold good is that the denominator
is controlled by nature, not by
consensus of people. So what would you
actually do as your third suggestion in
order to fix Let's not even talk about
America. Who cares? In order to fix it
to fix the over financialization that
you're talking about, would you try to
get everything back onto the gold
standard? Absolutely. At first, you have
to start. And by the way, you're not
going to the US dollar is going to zero.
Like the US dollar is going to zero. Why
is the dollar going to zero? What are
what are what are fake Roman dollars
worth today? Take a Roman dollar. If the
Roman dollar is a physical coin made of
gold, it's actually still very valuable
today. It's probably worth more than it
was in the Roman Empire. Well, so Rome
notoriously had a currency that it
eventually decided to devalue by mixing
the precious metal with cheaper metal so
that it could fund more different kinds
of at least there's still some precious
metal in there, right? At least the
denominator has gone up in like a clear
way. My point is even if you have that
Roman coin that's now only half gold,
it's still very valuable today. But if
you think about if you compare that gold
coin to like the the the the German yark
or whatever whatever fake currency
Germany had after World War I. the the
currency during the Vimar Republic. My
history is not that great. I'm not going
to I'm not going to I'm not going to say
that. But I don't need history to say if
your currency is made of paper and
backed by nothing, when your empire
falls, the value of that currency goes
to zero and all empires will fall. So,
but I don't think modern Americans
necessarily
even care that the dollar would go to
zero if America fails. They don't think
America is going to fail anytime soon. I
mean, sure, maybe it won't. But my point
is the US dollar is going to zero. We've
agreed on that. Now we're just arguing
about the time horizon. Well, humanity
is going to zero on some kind of time
horizon. Well, yeah. Okay. Well, the
dollar is not going to zero on the time
horizon of the heat death of the
universe here. Do you think that within
your in my lifetime the American dollar
is going to zero? I don't think so. I
think there's a decent chance. What What
would you put as the chances? Be a
betting man for a moment. 50/50 that the
US dollar goes to zero in our lifetime.
Us being mid-30s, we might live forever.
I might live a really long time. But
yeah, I think I think I think even if we
live to 100, I think Yeah. Wow. Yeah.
50/50 the dollar goes to zero. Yeah. All
right. And so what do you think the
right approach is if you think that
that's possible? So there's the biology
approach of gathering people up to start
their own network states. There's the
sort of practis approach of okay, can we
actually start a country and physically
colllocate people? What do you think
there is to do about this? I mean, I
think that's extreme. Why doesn't the
the US just issue a new currency backed
by gold? It exists in parallel with the
US dollar. We'll call it USG, US gold.
We'll make some like real nice looking
bills. We'll like hire some decent
graphic designers and make our money not
look like crap. Who's ever making the
money in Hong Kong looks real good. So
So we'll make some good-looking money.
that's backed by gold and you can start
using it. I think people would really
hesitate to I think governments would
really hesitate to reback their currency
in gold. Of course, because they've been
running a scam for so long. So, we got
to get rid of that government and put in
a new government that's going to build
decent currency. Here's another thing. I
mean, it's the same thing they're
already doing in India. You can have you
can have VUSG. VUSG is backed by real
gold, redeemable for real gold. Has lots
of warnings. You know, Google's thing
about like don't be evil, right? Like
like if I ever started a company, mine
would be don't be evil. And if we ever
remove don't be evil, it's because we
are evil now, right? Like that would be
the that would be the slogan, right? You
would say the quiet part out loud.
Exactly. So, uh you know, I'd make sure
to to put in a lot of friction. And it
is friction. It's always going to
fundamentally be friction that
says VUSG is is
uh redeemable for gold. And the minute
it's not redeemable for gold, this
currency is going to zero. You should
get out. Right? That's like the like
like that is printed on every bill. On
every bill it'll say this this bill is
backed by physical gold and redeemable
for physical gold at this address and
the day that it is not get out of this
currency because it's going to zero and
that's written on all the bills. You
want to enforce that as much as possible
and then I'll set up a alipe style
people can pay each other. You don't
actually have to carry around lumps of
gold or bills or anything with 0% fees.
And that thing would be run by the
government. It'd be run by private
companies. I think I think that we could
actually use blockchain technology for
that. Stable coins are perfect for it.
In fact, Franklin is building something
kind of I have no idea what the hell a
stable coin is. It's it's
it's not going to I'm not going to I'm
not going to fiat currency pegged. Yeah,
it's a fiat currency pegged. So, in
other words, it's pegged. It's pegged to
It's pegged to
on Yeah, it sounds
great. You know what? You know what? I
shouldn't have said the word blockchain.
We're going to use an openly
replicatable MySQL database that anybody
can verify all the transactions of. Or
maybe we'll put some ZK on top of it so
that they're private. We'll still store
the transactions in my SQL just so it's
not a blockchain.
How are you going to determine if this
MySQL techn is up to state? Well, we'll
have stakers. We'll have stakers who
attest to the state hash. Isn't that
just like a blockchain, dude? Are you
telling me blockchains were just my SQL
databases the whole time? No, because
they're much crappier than MySQL
databases. If someone said, "Here's a
MySQL database. It can handle 10
transactions per second." You're like,
"Dude, what is this? The 1950s make
currency that's backed by gold." And you
have a low friction 0% way for people to
pass it back and forth, right? You want,
oh, well, who's going to deal with
chargebacks? Yeah, that's literally the
job of a government. Who's going to pay
for this? That's literally something I
would love my tax dollars to go for a
welladministered 0% payment system for
everybody so nobody can rentseek on
that. Incredible. Let's talk about it's
not going to even cost that much. It can
cost nothing. Let's talk about compute.
Let's talk about compute, GPUs, CPUs.
Um, what do you think people don't
understand about Nvidia GPUs and the
market for different kinds of chips? I
don't think people understand anything
about it. I think people really have a
cargo cult understanding and they'll
just like repeat something and it
doesn't make any sense. People don't
know what CUDA is. you start with that.
It's like, let's start with what CUDA
is. It's actually funny. I tweeted that
no one knows what CUDA is. And then
Chris Latner replied and said, actually
at Modular, we know what CUDA is. And
I'm like, all right, Chris Latner, you
know what CUDA is. I'm not talking about
you. CUDA is an ecosystem. Said it's an
ecosystem for what? Like AI stuff. You
threw the AI stuff into CUDA. You use
the CUDA for the AI stuff. And what does
that have to do with Nvidia? Nvidia
makes the ecosystem. The CUDA ecosystem
includes Nvidia TM hardware. So a way
that you could describe this is that
CUDA enables Nvidia to have a moat
around its hardware and that if one were
to surmount the software moat around
Nvidia's hardware that their lock on the
the future might be disrupted. Yeah. So
the more like you have an ecosystem. If
something's the dominant ecosystem,
people want to build in the dominant
ecosystem. Mhm. So you have a lot of
people who build Yeah. You have network
effects. The people build in the Nvidia
CUDA ecosystem and it's not an open
ecosystem. AMD cards can't run CUDA.
Mhm. And like it's not even that
proprietary to Nvidia. Yes. And it's not
even that like AMD cards can't run CUDA.
That's kind of the wrong way of like
thinking about it. It's that there's a
lot of complexity inherent to CUDA that
make it only work well on Nvidia
hardware. Every couple years you'll get
someone who's like, "Oh, I built a
translation layer that will let me run
CUDA on AMD." And that's never going to
work. AMD is the number two competitor
behind Nvidia. Yeah. Because it's never
going to be fast. Oh, open CL. Like I
think all these things are kind of just
being done at the wrong abstraction
layer. But this is a specific tiny grad
bat. I think I think Chris Lightner
basically has the same bat saying that
instead of operating at this very
specific looking API layer right where
you have like something like you define
a kernel it has global size and local
size or CUDA's that is like that metal's
like that hips like that instead of that
you just say okay I have the tensor
compute and I have hardware to run the
tensor compute how do I best compile the
tensor compute for that hardware that's
what tiny bread is and so you think that
no company is going to be able to put a
moat around their compute hardware. I
think that companies are going to try.
Our mission statement at Tiny Corp is to
commoditize the pay to flop. What does
that mean? Our investors like George, do
do you really want to say that? I'm
like, yeah, I really want to say that.
When something's a commodity, there's no
uh rent seeking available. So none of
the stuff people are doing wrong with IP
and with their houses. Yeah. So, so for
the paflop to be commoditized and what
that actually means is there's many
chips that the idea of a payoff flop the
idea of a chip that can do a floating
point calculation is obviously a
commodity. There's many chips you can
buy that can do floating point
calculations. Nvidia GPUs, AMD GPUs,
even Intel's things are capable of doing
floatingoint calculations. Many many of
them. Can you give people a sense of how
much compute one pa flop is? Like how
much compute in an iPhone? How much
compute in a data center? So pa is 10^
the 15. Um, an iPhone is on the order of
10 teraflops. So, a 100 iPhones is one
payoff. Got it. So, maybe a 100
teraflops in a new iPhone, 100 tops. So,
maybe it's 10 iPhones, but somewhere
between 10 and 100 iPhones.
And so, if you succeed in that mission
and the pay to flop becomes a commodity,
what does the world look like? So,
currently there's an extreme premium for
Nvidia flops over other flops. When you
look at the 4090 versus the 7900 XTX,
they're very similar looking cards. They
have very similar natural hardware. Not
I mean there's like there's nuance and
details here. But my point is per flop
there's a 2 to 3x premium for Nvidia
hardware because of the network effect
because of CUDA because of the moat.
Yes. So there's that 2 to 3x premium and
then there's another 2 to 3x premium
which is a technical premium where
people buy these data center cards that
use like HBM memory because has higher
bandwidth and lower power but I don't
think you need that either. I think what
you basically want is
a huge number of racks of consumer GPUs
uh running your models. So just to to be
clear there are two kinds of GPUs that
are being sold. There are kinds that are
intended to be put in data centers and
there are kinds that are sold directly
to the consumer. and these two things
have different margins. It's not all
about margins. There is a margin aspect
to it, but there's also just some
technical decisions being made in the
data center GPUs that make them more
expensive. So, when you look at just a
flop, if you think about a flop on an
AMD consumer card versus a flop on an
Nvidia data center card, there's a 10x
price difference in that flop. And that
10x is what I want to commoditize. Mhm.
And then also I think it could be more
like a 100x if there was more of a
market down here. There's not that much
of a market for consumer AMD flops.
There's a huge market for Nvidia data
center flops. If there's more of a
market down here for just like any kind
of flops, all flops are like commoditiz.
Every flop should be equivalent. That's
what a commodity really is. Compare a
fungeible token to a nonfgeible token. I
don't I want flops to be funible. I want
every flop to be created equal. That's
the dream. That's the dream of the tiny.
All flops are created equal. And Tiny
allows you to use those flops to do
useful compute. And so you just started
using some flops from AMD. Been using
AMD for a long time. AMD finally came
around and agreed to give me some
computers. Okay. Again, I can buy the
computers, but I like to I like to test
for buyin. are you interested in what
I'm doing or not? Thank you for the
computers, AMD. A good cultural test and
I think that AMD passing my cultural
test means that they're understanding
the value of this kind of stuff and I
see a bright future for them. So you
have this external culture test but also
you are running a very unique internal
culture around how tiny is organized and
managed. Maybe you can talk a little bit
about that. Um I don't think that it's
like that radical anymore. I think
everything is basically going to go to
this format. So, we'll start with like
we don't do interviews. Your resume goes
right in the trash. God, you went to
college. Yeah, I don't care about that.
The only way that we hire is from people
who've contributed to the tiny grad
repo. Mhm. And then we have bounties.
So, you can get paid right away. And if
you are really skilled, you can make we
we had one guy who's just like tearing
through the bounties and he was like
making two grand a week just tearing
through things basically and like not
even full-time. If you are very skilled,
you can get paid a full-time salary off
of bounties alone. Are these folks
contributing to your bounties? Are they
themselves using AI for coding? The
thing about using AI for coding is AI is
below the bar of the person we would
hire at Tiny Corp. So, some companies
may have a very low bar. If you have a
very low bar and you you like if you
have employees that are worse coders
than chat GPT, oh god, how did you hire
them? What was your process here? uh
they should all be fired and they should
all be replaced by judgment team. So we
encourage everybody to use AI to the
fullest of their ability. This isn't
like we're not going to like oh you
can't use AI for the interview. No, use
AI. Please use AI. Please use AI when
you work here. The only thing that I
care about at the end of the day is the
final product. And I think that
everything is going to move. It's a
machine learning concept. You have a
training set and a test set. And you
want your training set and your test set
to come from the same distribution. If
your interview is your training set and
the actual work is the test set, you
want your interview and your work to be
as close as possible together. Think
about it. If you're a fang company and
you do leak code interviews and then
what you actually have to do is fix this
bug in this web endpoint for people in
Nigeria trying to pay in Empa like
there's just there's such a disconnect
there. So that disconnect should go to
zero. And that's kind of the first
premise of of Tiny Corp. The only way to
get hired here is to contribute to tiny
grad and then once you do get hired
here, it's the same job. Are there
specific tools that you feel like are
better and worse for assisting these top
level coders currently? No, it's it's
all the same knowledge as everybody
else. I think that they're You don't
have a preference amongst them. I think
they're good for different things. Um,
Chad JPT is probably the all-around
best. Uh, Grock is recent. It's got
great recency, but it's prone to
hallucination. Uh, and then Deep Seek is
the best at like thinking. If you want
something that's really going to think
through something, JBT is kind of lazy.
Deep Seek works hard.
Uh, maybe something to be said there
about America and China, but uh, yeah.
So, yeah, I think I think I use some
combination of Deep Seek, Chad GPD, and
Grock. Same with everybody else. One of
the risks that you and I have talked
about with AI and of course you're not a
doomer in the Udkowski sense is what
you've referred to as wireheading. What
is that risk and do you see that
happening now? So wireheading is simply
when you've hijacked your own reward
function. So your brain has some reward
function. Your brain has things that it
finds pleasurable, like food, like sex,
like productive work hopefully. But it
unfortunately also has other things that
it finds pleasurable, like Tik Tok and
Twitter and fentinel and meth. Uh, and
these are unfortunate because they're
nonproductive. They're these they're
these non-productive ways that stimulate
your your your pleasure center. I don't
think they're all like truly terrible
things and in moderation, whatever. But
the the term wireheading comes from this
experiment done on rats and they drill a
little hole in the back of the rat's
brain uh bring little little electrode
right into the brain stem and uh the
electrode would would pulse a little bit
of electricity deep down in the brain
where your pleasure center is and hooked
it up to a button and the button gives
you a little electricity and the rat's
like, "Oh, button. Oh, button. Yeah,
button. button. Oh, button button
button. And and the rats die and they
don't die because of a overdose of
pleasure or anything like this. They die
because why would you eat when you could
press button? Entertaining ourselves to
death. Entertaining ourselves to death.
Amusing ourselves to death. I think
that's the that's the Neil Postman,
right? Uh yeah. So I think that this is
that this is definitely happening and
that we will
lose some chunk of humanity to it. Do
you see it happening already? Yeah, but
not to the extremes which are well. I
was talking to a 21-year-old who just
graduated from a great school and she
told me that a big chunk of her
classmates are dysfunctionally addicted
to Tik Tok. As in, they live with their
parents. They can't get out of bed and
do stuff. They aren't looking for jobs.
They're just watching Tik Tok all day.
Do you think we're starting to head in
that direction? I'm always reluctant to
you can go back to like the quotes where
Socrates said, "Wow, the youth of today
is is so is so corrupt and obsessed with
there's always that article, right?
There's always that article of the older
people saying the kids are not all
right." Yeah. So, I'm not so sure that
we've really even seen it yet. Um, and
I'll get to what it might sort of look
like. And then people like, "Oh,
fentanyl. People are opiate addicts."
Yeah, but like then you go back to like
the opium wars a 100 years ago. I'm not
so sure this stuff was that different. I
mean, it's stronger. Yeah, but I don't
know. It's all It's all interpreted
through a cultural lens, right? There's
no problem with Fentel in China today.
Well, there's no crisis of meaning and
purpose. Yeah. So, I'm not so sure that
these things even Tik Tok are are
reflective of it. But there's I mean,
there's a great story. It's uh it's uh
My Little Pony friendship is optimal. Uh
and I'm not going to I'm not going to
give away the You're going to out
yourself as a brony. Oh, I mean I have
watched many have watched many episodes
of My Little Pony: Friendship is Magic.
I can name the core six, uh, you know.
So, yeah. Um, I can name the different
types of ponies. Yeah, I do. So, the
rationalist version. Yeah. And and I I
think that that wire heading is going to
look a lot more like that. Or we can
talk about the concept of heaven
banning. Heaven banning. It's some guy
on on Twitter near Sion uh popularized
it and it's an interesting idea where
heaven banning is this imagine you're
posting on uh social media you're
getting lots of replies praising you
giving you the kind of feedback you're
looking for giving you likes giving you
retweets but of course all of these
people you're interacting with are AI
bots I don't think that wireheading is
possible without sufficiently advanced
intelligence I don't think that humans I
mean maybe if you do it like with a
drill and an electrode, sure you do a
drill and an electrode will bypass
anything. But if you're trying to do
wireheading through an app, I don't
think wireheading through an app is
going to be possible until that app has
sufficiently more intelligence than you.
What are you doing personally today
yourself to protect yourself from
wireheading and various kinds of amusing
oursel to death addictions? There's many
you you've asked several of these sort
of questions like
They're like alpha mining. And I feel
that there's way too much of this in the
world. Like way too much of these like,
"Oh, can you tell me the secret to being
a good programmer?" And I tell people
the story. You know how I became a good
programmer? I'm working on it. Oh, no,
no, no, no, no. So, there's a
volcano. It's a true story. Uh, so, so,
so in in in uh in Indonesia, so you
know, there's the island with Bali. Uh,
you can go all the way to the west of
that island, and then you can get on a
boat, and that boat will take you to the
main island. Mhm. And in the main
island, there's this city called Ban
Yuanji. No one's heard of this city.
Maybe you've heard of Jakarta, maybe
you've heard of Surupaya, but no one's
heard of Ban Yuanji. Okay. And then an
hour outside of Banuangi, there is this
volcano called Aenet. And this volcano
is the only volcano in the world where
there is blue flames. There's sulfur
released by the volcano. The sulfur is
lit on fire and it turns blue. So one
night I set out in a
car, went to the volcano. You have to
climb up the volcano. I rented a gas
mask from an Indonesian man. Just true
story. I put that I put the gas mask on
and I I I climbed down to the bottom of
the volcano and I saw the blue flames
and at that point I understood
programming. Okay. So just to defend my
question because I feel like it it
requires defending. People like Cal
Newport, Digital Minimalism, have
written books with the intention of
helping people uh protect themselves
against companies spending billions of
dollars with data scientists to
monopolize their attention and change
how they think. You have no such
recommendations to offer and think it's
an illegitimate exercise to make such
recommendations. Is that the same guy
who made Newport cigarettes?
I don't think so. I I'm sure they were
there to help people, too. Big family.
Big family. Lots of different or you can
read seven habits there. Newports highly
effective people or how to make friends
and influence people or the 4-hour work
week or you can read. And do you think
those things are valueless? Yeah, of
course. You don't think people picking
up habits by mimisis is useful? I mean,
I don't think they're like valueless.
Like they're fine books to read. They're
not like, "Oh god, you read that book.
It's a terrible book and it'll lead you
in the wrong way." But no, I I don't
think that there are I think that a lot
of people spend way too much time
thinking about what I should do or how I
should optimize things and they should
actually just do things. I think that
people spend a lot of time in
transances, failing to introspect about
how they're thinking, how they're
spending their time. And I think helping
people break that trance enables people
to live more self-directed lives instead
of falling victim to companies that just
want their attention for money. I'm an
emo kid. Non-conforming as can be. You'd
be non-conforming, too, if you look just
like me.
All right. Well, is that a good place to
leave it? Those are my questions. Thanks
for being on, George Hots. I appreciate
it. Bye, Amanda Show. Watch the other
episodes. Watch my Twitch stream. Don't
watch my Twitch stream. Thank you for
listening to this episode of Endgame.
Please follow me on X, Amanda Casset.
Two S's, two T's. Follow the company I
founded, Serotonin HQ, and listen
wherever you get your podcast.
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
Works with YouTube, Coursera, Udemy and more educational platforms
Get Instant Transcripts: Just Edit the Domain in Your Address Bar!
YouTube
←
→
↻
https://www.youtube.com/watch?v=UF8uR6Z6KLc
YoutubeToText
←
→
↻
https://youtubetotext.net/watch?v=UF8uR6Z6KLc