---
title: 'The AI Race Just Changed Direction'
source: 'https://youtube.com/watch?v=Al04Qrn03bo'
video_id: 'Al04Qrn03bo'
date: 2026-06-21
duration_sec: 2052
---

# The AI Race Just Changed Direction

> Source: [The AI Race Just Changed Direction](https://youtube.com/watch?v=Al04Qrn03bo)

## Summary

The video discusses major shifts in the AI landscape, including China's restructuring of its university system to focus on AI, new AI memory systems that can track lost objects, and the implications of AI on various industries. It also covers topics like in-game advertising by Electronic Arts, the need for AI safety agreements, and the changing market share of AI models like ChatGPT.

### Key Points

- **China's University Overhaul** [0:00] — China is cutting over 12,000 degree programs and redesigning its education system to focus on AI and STEM fields, dropping arts and language programs to align with national development goals.
- **AI Memory System for Lost Objects** [0:11] — MIT researchers developed a system called DAAM (Describe Anything, Anywhere, Anytime, Any Moment) that helps robots remember where objects were last seen by building a 3D map with rich labels.
- **EA's In-Game Advertising** [7:26] — Electronic Arts launched EA Advertising, allowing brands to appear inside games on stadium signs, billboards, and custom items, reaching over 120 million players monthly.
- **AI Landscape and Government Response** [10:08] — AI is moving fast, with governments and banks showing cracks. Anthropic published a plan threatening mass job losses, while the US government uses national security powers to control AI access without clear legislation.
- **Humanity Not Ready for Intelligence Explosion** [12:22] — An article argues that AI is advancing faster than governments can prepare, comparing AI risk to nuclear risk, and calls for serious AI safety agreements based on verification and inspections.
- **Tim Ferriss on AI and Book Publishing** [16:19] — Tim Ferriss argues that AI is collapsing the market for how-to non-fiction because people want answers, not the path. His print catalog is down 80% compared to 2022.
- **Sharing Algorithm: Humanity Should Benefit from AI** [18:59] — David Kurpatre argues that since AI is built on human data, the wealth it creates should benefit all of humanity, not just the companies that built the systems.
- **Humans Outperform AI on Math Test** [21:09] — In a rigorous math test with unpublished research-level problems, AI systems solved up to 6 out of 10 problems, but humans still outperformed them, showing AI is not yet a full replacement for human experts.
- **Harsh Critics Are Most Loyal Fans** [22:35] — A study of 23,000 Reddit comments found that harsh critics of video game brands are often the most loyal fans, complaining because they are invested and want improvement.
- **Cyber Experts Warn Fable 5 Ban Could Backfire** [24:06] — Over 100 cyber security experts argue that banning Anthropic's Fable models weakens US AI and gives advantage to attackers, as open-source models can already do similar work.
- **Global Opinion Shifts: China Leading in AI** [25:27] — A poll in 15 countries found that in 11, respondents believe China has passed the US in AI capability, with only 51% of Americans saying the US is leading.
- **ChatGPT Market Share Drops Below 50%** [27:17] — ChatGPT's market share is falling due to competition from Gemini, Claude, Meta AI, SpaceX AI, and Chinese models, with OpenAI under pressure from all sides.
- **Codex, Cursor, and AI Competition** [29:08] — SpaceX's $60 billion acquisition of Cursor is seen as a strategic move to dominate coding AI, which is considered the flywheel for all other AI capabilities.

### Conclusion

The video highlights that the AI race is intensifying with China restructuring education, new AI memory systems, and in-game advertising. It emphasizes the need for AI safety agreements and notes that while AI is advancing rapidly, humans still outperform in some areas, and global trust is shifting towards China.

## Transcript

China's universities have cut more than
12,000 degree programs and they are
redesigning their entire school system
for artificial intelligence focused
majors. We're going to look at what
people call the holy grail of artificial
intelligence breakthroughs, telling you
where you lost your keys. Electronic
Arts, the game company, actually
released a new way for for you to
advertise directly into video games. I
was like, this is like another Google,
right? I mean, if they build the
plumbing for an entire ad marketplace
that lives in three-dimensional space, I
was like, I'm looking into it. We'll
look at some thoughts from Rob Mason,
take the 10,000 foot view, step back
from the world of technology today, and
figure out where are we, where are we
going, and where have we been. I'll
break down the viral economist post
about how humanity is not ready for the
upcoming intelligence explosion. I'll
kind of share some of my thoughts on
what they wrote. Author Tim Ferris, who
I like I remember reading his books like
decades ago, and I kind of forgot. I
know he's been a podcaster for a while,
but he just wrote this thing about like
what AI is doing to the book publishing
world. Then we'll get a little more
philosophical about who should be
benefiting from all of this change.
David Kurpatre wrote the sharing
algorithm. If AI is built on humanity,
shouldn't humanity benefit? I'll share
kind of the flip side of some of all of
these AI tests that I'm always sharing
because there was a really stringent
math test and humans actually won it. AI
did not take the crown even though we
see all of these breakthroughs all the
time. I mean, I don't know if they will
be forever or how much longer, but today
humans win. Some internet psychology
that took me by surprise. Turns out
harsh critics of things are usually the
most loyal fans, especially of video
game brands. But it actually kind of
gives me a different perspective if
they're actually the most loyal people.
So over 100 cyber security experts are
saying that the Fable 5 ban could
backfire on the US. But experts are
growing around one consensus and that is
China is starting to crush the US in
terms of AI growth. I mean there's a lot
of ways to look at it. Certainly America
has the lead in some very serious ways,
but it's kind of wild how many edge
cases we're starting to lose, especially
when we have so much money and we had
the head start. You know, good old chat
GPT is now below 50% of the market
share. And actually, after we talk about
that, I'll dive a little bit into the
big business of what's going on with
this SpaceX IPO. It's it's a wild time
to watch billionaires jockey for
position. And there's it's pretty
interesting positioning. And hey, when
you're a trillionaire, it's like not
even a tenth of your net worth. And you
don't even have to pay for it. You just
use your companies that have trillions
of dollars. So, I get it. Just buy it. I
mean, unless of course you care about
citizens having any kind of say or
control in our democracy, which nobody
does.
But look, Anthropic figured out the
playbook about why coding is the single
focus that you should be having and
everyone else got a little distracted
with side quests. So, we'll see if now
that everybody sees what Anthropic sees
where it all plays out. But first, if
you don't mind heading over to Dylan
Curious, actually, you're probably
already there if you're watching this on
YouTube, and hitting the share button,
just like boom, share, copy, and send it
to anyone. That would be super helpful.
I know you guys did that on the last
video, so thanks. It definitely helped.
Look, you can see it did above average.
Like by day four, we almost hit 10,000
views. I made $77 on that video. That's
crazy. I'm going to get some food.
That's like four or five Chipotles. All
right, so what are your thoughts? China
is now cutting thousands of university
degrees. Didn't even know there were
that many. But they're thinking about an
AI era and it kind of makes sense. I
mean, it's a lot of government control
over school, but also like we kind of
need to get a little kick in the butt
here to get ready for these huge changes
that are coming. China's universities
are making major changes to what
students can study. The country is
trying to line higher education up with
a national development goal and that is
to dominate and win against the United
States in the future of artificial
intelligence and technology and many
other STEM programs. Arts and language
programs are being dropped while tech
focused fields are being favored. And
it's what universities should do, right?
They want to be preparing students for
industries that are most important to
the next decade. So that requires a
national reshuffleling of education amid
the backdrop of a much more techfocused
future. All right. So if you've ever
misplaced anything, which I think is
pretty much all of us, I don't know how
many do actually we should put in the
comments like how many times a week do
you lose something like your phone?
Anything like phone or like scissors if
I'm trying to like open some boxes or
something. I probably misplace things
like a dozen times a week. Yeah. Like a
couple times a day or something. And it
kind of sucks. I'm like or especially my
AirPods. Actually, probably maybe more
than that. I'm always using the find my
feature where I'm like beep beep beep
and I'm like oh it's on that side of the
house. Yeah, that's right. But listen,
people lose stuff all the time,
especially groups of people working on
construction sites or big companies like
it can just get confusing. You're
focused on your work, you set something
down, someone else uses it, whatever. If
you think about it, AI with camera
vision watching everyone should be
easily able to just be queried. Hey,
where did I leave my phone? If that
information is in context, it will
remember when and where it was lost. And
even if it's just visually available, it
should be able to just, you know, hunt
it down. So researchers are building
robots that may soon remember places
more like humans do, including where an
object was last seen. And you can jump
into the eyes of the AI to see the same
thing that you should have remembered.
So a worker can remember where she or he
left a part in a factory last night. And
a robot can do the same. A robot usually
cannot do that well at this job because
it struggles with memory that connects
places and times together. But MIT
researchers have built a new system to
help robots learn that kind of memory.
It is called DAM. But it's got extra A's
in there. So it's not a swear word. It's
D A A M. It stands for describe
anything, anywhere, anytime, any moment.
Oh my gosh. Actually, the acronym is
missing an A. It's D A A M. Wait, what?
So 1 A is anything. Second A is
anywhere. Third one is any time. And
then at any moment. You could have you
could have thrown a fourth one in there.
They missed an opportunity there. But
anyways, as the robot moves through a
space, it builds a 3D map. But it also
adds rich labels to what it sees as it's
going along. For example, it could
remember a bike rack outside of a campus
building. Maybe it has five bikes on it,
one red, one yellow, one with a flat
tire, etc. The system then groups
objects by location so the robot can
answer in plain language where something
is later. And that is a new AI system
where you can ask a question like where
is the part that we started assembling
last night. You don't even have to have
it be something that it could tag like a
part that doesn't even have a specific
name because it can figure out what you
mean based on the context of it. Knowing
it was last night, knowing it was
something you were assembling and it can
just eliminate all of those other
choices. And it has a special system for
choosing clear images, describing
several objects at once and avoiding
labels on the same object again and
again. And that actually was something
that was really hindering some of the
systems before this. So, what do you
think? Do you think that's too much
privacy invasion or do you just like the
idea of knowing an AI knows everything
about you and where you lost everything?
All right, let's talk about this new EA
advertising thing. Like maybe if you're
already in the game world, this isn't
like such a surprise to you. Maybe
there's been like ads kind of shoved
down your throats for a long time, but I
don't really play much games, so I was
not ready for this. I always imagined EA
was just going to make money based on
the games that it sold. I kind of knew
it was moving towards a world where
there's like recurring fees or maybe
they like, you know, charge you for skin
upgrades or whatever, but EA is going to
get into the ad business. Like now
they're going to fight for your
attention inside of a game. Like what?
So Electronic Arts has launched EA
advertising. This is a new system that
lets brands appear directly inside its
games. That means ads can show up on
stadium signs, digital billboards,
scoreboards, broadcasting graphics,
custom in-game items, and branded
challenges. Is this going to ruin games
or is this okay or is it make games
better be free at least? The idea is to
make ads feel more like part of the game
setting in the same way brands already
appear in real sports broadcasts. He
says it reaches more than 120 million
players each month. Oh my god, that's
like the entire Super Bowl audience. It
also says players are spending huge
amounts of time in games like Madden,
the NFL, EA Sports, where matches,
seasons, and live sports style moments
happen consistently. Like, you'd think
if the NFL licenses the entire NFL to EA
to make a game out of, wouldn't the NFL
still own the brands or can EA just do
this? And what does this mean for the
future of VR and AR where everything's
just being automatically generated? But
yeah, this the kind of architecture
behind it already includes reward-based
goals, vanity items, branded content,
live events, creator tools, social blade
features, and community programs. What
EA was the company that was just bought?
It was bought, right, for like $60
billion or something? Was EA recently
bought? And if so, who was the buyer and
price? What was the logic? Yeah. Okay,
so $55 billion leveraged buyout, Saudi
Arabia's sovereign wealth fund. That's
right. I remember talking about that. I
wonder if this was their plan. like they
were just thinking, gosh, it's going to
be so valuable if we can take some of
Google's profits from ads and get them
into games. Honestly, we're like later
in this video, we'll talk about SpaceX
buying Cursor for 60 billion. Th it
would not shock me at all if SpaceX
purchased EA for this much money because
I mean, he's talked about creating a
video game company. And if you already
have Cursor and you already have like a
bunch of inference systems that you can
take back from Anthropic anytime you
want, you know what? That might honestly
happen. Maybe EA implements this ad
thing, generates some insane revenues,
and then sells it for what do you think?
$180 billion to SpaceX in two years.
There you go. That's my prediction. Let
me know in the comments if you if you
want to be on board with it. Over,
under, happen, not happen, true, false,
however you want to bet. All right,
let's look at mid June 2026. Where are
we at right now in the landscape of AI?
Because you I mean, you know, if you
watch this video how fast it feels like
it's been going recently. Um, AI is
moving fast, right? And I think I you
see cracks all over governments and like
banks. I think young voters are now
really thinking about it like it comes
up sometimes with politicians in a way
that it never used to. And Rob Mason
kind of broke this all down for us. So
in one major a AI lab, Anthropic to be
precise, they published a plan that
threatened mass job losses as a real
possibility. We talked about this in the
last video too. Universal basic income,
public wealth funds, some of those
things. Um, timing was obviously messy
because they're also trying to go IPO
and they want to look like super
profitable and super capitalistic at
that time. So that, you know, cha-ching.
So critics were asking whether this was
real policy support or just smart IPO
positioning, basically being like, we're
going to gobble up all that value.
Invest now. And then few days later,
government says, no more Fable 5. Take
away your like frontier model. Here's
where we're at. The government sometimes
is like totally handsoff. We just want,
you know, business to like go. It feels
like kind of a corrupt government.
nobody's going to stop anything and like
just big companies and K-shaped
economies are going to happen. So, the
government has sounded hands off on AI,
but then it's used national security
power to control access without clear
legislation with all of this mythos
stuff. Europe and the UK are reacting by
taking it more seriously, but they're
talking about sovereign AI, meaning they
want to control their own advanced AI
systems and they're trying to keep money
inside their own borders. There's
definitely a lot of fences going up all
over the world. And at the same time, US
senators, major banks, young voters,
they're all moving into the debate in
different ways. And some companies are
just openly talking about AI replacing
roles through automation. And it's just
it, you know, it is their fiduciary
duty, I suppose, but it also just
doesn't resonate well with a society
that's coming into the workforce. Young
people are becoming less excited and
more angry about AI. So, we'll have to
see what happens. So just on top of like
where politics and technology are
intersecting, there's also just a
natural trend, right? That this is like
Ray Curtzwhile has been talking about it
forever and we're getting into the
hockey stick part of it. So this article
is called humanity isn't ready for the
coming intelligence explosion and it's a
classic talking point on this channel.
Uh AI leaders and everyone who's
enabling them are moving humanity
towards a race that is making smarter
and smarter machines. And it's doing
that without a real serious good
trustworthy safety plan around it. Tons
of emergent properties, tons of things
that can't be predicted when systems are
actually live and in people's hands,
tons of things that are completely un
unpredictable when different agents are
acting with different intentions and
different goals in some kind of like
bigger system. So could AI be advancing,
becoming more dangerous faster than
governments are ready for? Absolutely.
That's probably likely, not just a
possibility. The article is comparing AI
risk to nuclear risk. Nuclear plant is
expected to have a tiny chance of a
disaster. You know, a lot of people
disagree with it and there are smart
people that disagree with it. So, I
wouldn't, you know, bank on it. That's
just what feels right to me. And I'm
sure a lot of new technology felt a lot
scarier than it was when people first
heard about it, before it was
implemented, before structure came. So,
it might move down. But, you know, the
concern is not just coming from
outsiders. It's coming from industry
insiders too. Some of the people inside
the biggest labs especially like
recursive self-improvement. I mean you
can see Anthropic and Musk moving
towards it at the same time worried
about whatever like happens when
self-improvement is AI writing its own
next model then generating it and then
it writing its own code. I mean this is
why a $60 billion investment in cursor
might be nothing to SpaceX in the long
run when it's a you know hundred
trillion dollar broke the economy kind
of company. And once there's an
intelligence explosion, if the
guardrails aren't good, there might be
no way back. How are you going to, you
know, there was that there's this
metaphor once where somebody said, "Go
up on the top of a hill and like cut
open a pillow of feathers and like let
the wind carry those feathers away and
like now go put them all back in in the
pillow." Like you can open it up pretty
easily. You can just take a knife to it,
cut it open, let all those feathers out.
But to actually like put them back
together, it's a much harder problem.
and evolution and and growth like this
now like actual intelligence in the way
we see it in the biological world it
once it gets going you don't just go
back so the economist is now arguing
that America and China in particular
because these are the two countries
controlling the vast majority of this
progress and need to create serious AI
safety agreements right now not based on
trust based on verification based on
inspections based on clear red lines if
you go a little too far and slow down
progress so be it like we'll have plenty
of time when these things are
exponentially growing to reap the
benefits, but really right now
biological weapons, cyber attack, fraud,
child abuse, like there's child abuse
material like being generated like all
this stuff is coming really fast. So the
article ends with a warning that
civilization may fail when powerful
tools grow faster than their ability to
govern them. What do you think? Let me
know in the comments. Yeah, I mean it's
hard for me to say pause, right? Like
I've never quite been a pause AI guy.
Although, if I could magically just stop
everything right now, I would. I just
don't really think of any realistic way
to just tell everybody like pause and
have important people pause and have
that be a a good outcome. But I kind of
do agree that maybe just big governments
right now getting serious about like red
lines might and inspection like all
getting all that system in place for
inspections might be a really smart
move. Now, I know bureaucracy is
bureaucracy, but like let a bunch of
people see what's going on behind closed
doors and like force their hands as much
as you can. Like one good thing about
China and America, if we both let each
other kind of look and see, there would
be stealing and there would be copying,
but on the other hand, we'd be seeing,
you know what I mean? Like those
conversations would get out and when
super dangerous things happen, everybody
might be aware of it at least. I go talk
to Tim Ferrris, 4-hour work week. Like I
remember he was he was like a in my
world like a little cultural phenomenon
in my a couple decades ago. I haven't
heard much from him but he actually
mentions the 4-hour work week which was
like his self-help book from a long time
ago and how it's been selling and how
the industry is getting crushed. So
having him talk about AI I found sort of
surprising. So his core argument is that
AI may be collapsing the market for how
to non-fiction because people
increasingly want the answer not the
path there. And that's interesting.
Like, is that another byproduct of using
these LLMs all the time? So, if a book's
main value is to tell me all the steps
that I need to take and a chatbot can
now feel faster, cheaper, more
personalized, and more convenient,
what's the point of self-help books? But
Tim Ferris is also arguing that the real
separation is between information and
transformation. So, information is
something like, give me five steps to
lose fat, but transformation is walk me
through a carefully designed journey
that makes me actually change. And that
difference matters. Chat bots can spit
out fat loss protocols in 15 seconds.
But Tim's example is when he took the
4-hour work week, which is this whole
body transformation thing, and distilled
it into some bullets and then sent it to
friends, none of them acted on it. But
thousands of readers who followed the
full book's journey felt what they
needed to like incorporate the
information deep enough to actually
change their lives. So, when I read this
and then I started using LMS like I
always do to um I do like moving some
stuff around and trying to figure out
how to like set up everything. I was at
the end thinking did I actually learn
how to do this for someone else? And I
thought to myself like kind of not right
like I if if someone else asked can you
help me with this? I would say like yeah
bring up bring up chat GPT and then ask
this question and then it'll walk you
through the steps. I think we went here
first and I would follow it again. and I
wouldn't actually be able to tell them.
So, I didn't process it and it kind of
freaked me out. And he's saying that
prescriptive non-fiction is getting hit
really hard. So, he says that his own
print catalog may be down about 80%
compared to the 2022 year. So, there's
clearly less people just buying books.
But self-help, he thinks, is especially
vulnerable. Eventually, the same threat
is going to be extended to YouTube, like
the videos I make. like you might might
not get as much out of this whole thing
as you could just like go down to Gemini
and distill this into like five news
bullet points or interesting thoughts
and then move along, you know, like I
understand maybe some of you would want
to do that. Podcast, courses,
newsletters, advice blogs, like I guess
if you just want to extract actionable
information instantly, that's a better
route to go. Next up, let's talk about
the sharing algorithm. If AI is built on
humanity, shouldn't humanity all benefit
David Kurpatre? So he's arguing that AI
didn't really come out of nowhere,
right? Like artificial intelligence was
built from human language, books, music,
code art research news questions
jokes, years of public investment, lots
of people posting things on social
media. But then companies built powerful
AI systems and all of a sudden that work
mattered for training. So if AI creates
huge wealth, should only the companies
that invented the thing that sucked it
up and learned from it benefit or should
all of us? I mean, the idea that a
shared resource creates wealth that the
public can own seems like one of the
kind of futures that we almost
inevitably will fall into. Like humans,
the interesting things humans do, the
data that we make, like that could be
pretty valuable. I do think synthetic
data is actually going to make
everything people do kind of, you know,
not as important as we'd hope. I I don't
know how long human data will matter,
but it probably will always matter.
maybe at least enough that companies
could somewhat pay taxes or the
government would be rich enough or
something that there could be some kind
of universal basic income. And I will
say for a long time I actually thought
what's going to happen to AI is they're
going to digest the internet like OpenAI
was the first to kind of grab it and
then I thought maybe right after them
Google and a couple others and then
they're going to produce so much garbage
on the internet from chat bots that
there won't be anything to put into the
system. But over time I have come to
just think that oh synthetic data
actually I think it actually works.
There's different novel ways to pull
data off the internet. There's different
ways to generate language in a way where
it seems like the systems do learn. Um
in some cases maybe they do kind of move
to the middle and don't like have that
massive creativity. But also I don't
know. I got a feeling I got a feeling
that's not going to be as as obvious as
I once thought.
Especially
with like physics simulations and stuff
like that and diffusion models of just
like oh okay you can you can actually
kind of come up with some pretty unique
random stuff to learn from over even
though you generated it. It's weird but
it seems to work. You just kind of
iterate towards better and better. But
in some cases humans still matter. How
about this one? Humans outperform AI on
this highly rigorous mathematical test.
This is good. I thought we were going to
get crushed after like reading all these
like math olympiad was destroyed by open
AI and by deep mind. But a project
called first proof gave AI models a very
hard math exam. This was not the kind of
school math. This was 10 research level
problems from mathematicians own work.
The problems were completely unpublished
at the time of testing. So the models
could not have just seen the answer and
repeated them. 30 mathematicians checked
their answers and that made this test
very different from any of the earlier
AI math demos that have ever been done
before. This was new work. It was done
without human help and it was judged by
real experts. Four AI systems took this
test. OpenAI entered Chat GPT 5.5 Pro
and then the university teams built
automated systems called harnesses on
top of chat bots like Gemini and Claude
for testing. Also, the harnesses asked
the models questions. check the answers
and pushed to improve. Now, the best
system was chat GPT. It was GBT 55. It
solved six of the 10 problems. It did
not match human top mathematicians. The
results showed AI can help with serious
math, but not yet a full replacement for
human expert problem solving. Wuzzing.
What do you guys think about that one?
Humans still in the lead. All right.
also had a new epiphany that some of the
harshest critics potentially even not
just in games but like in my comment
section might be the most loyal fans. I
thought I thought the harshest people
were just I was like gez guys like just
chill. But no, harsh critics are the
most loyal fans of video game brands.
Study shows a new study looked at the
way video game fans stay loyal to brands
like Call of Duty and Battlefield. And
instead of bringing people into a lab,
the researchers used AI to study more
than 23,000 Reddit comments. They wanted
to see how loyalty shows up in real fan
communities. And the results were not
what people expected. There wasn't one
clear pattern either. So, Call of Duty
fans showed showed loyalty through
emotion, stories, and shared memories.
But Battlefield fans showed it through
skill, technical details, and respect
for how the game works. I was like,
"Wow, these are really deep communities,
like different languages, same deep
loyalty." And the most interesting
finding is that harsh criticism is
usually a sign of commitment. Fans do
not always complain because they hate
the brand. They complain because they're
invested and they want the game to
improve. The same person who tears apart
a flaw may defend the broad movement in
an outsider attack. So the researchers
are telling brands like you should stop
trying to silence useful criticism. Like
they're always trying to cut. Remember
the uh was that Cyberpunk that release?
Everyone's like just tearing on it. like
those people probably really wanted the
game to be good. They also learned that
like fake corporate language like really
pisses off fans. So maybe try to be a
little more authentic when you talk to
them. But either way, food for thought.
So let me know in the comments. Do you
think that the um mythos uh pullback was
actually good for the United States in
their position in the world? Do you
think it's good for companies to not
have cyber threats or do you feel like
um it actually made us weaker? because
there's some debate could go either way.
But there's this open letter signed by
hundreds of cyber security experts that
asks US officials to lift export
controls on anthropics fable and mythos
AI models. The writers say that AI is
already changing cyber security and it
can help find software flaws and write
exploits which are the tools that attack
those flaws. But they argue that
anthropics models are not uniquely
dangerous. The major and open source
models can already do similar work.
their concern is that taking strong AI
tools away from the defenders gives an
advantage to the attackers or do you
think that the attackers not having it
make them weaker? You know what I mean?
It's so tricky. They're arguing that
security teams need these models to find
and fix bugs in new code and old systems
before the adversaries do. The letter
also says that Fable already had strict
protections against offensive cyber use.
So banning access in total in their view
just creates the market fear and weakens
American AI without leading to any clear
risk reduction. So consensus is growing
that China is crushing the US at AI.
There is a few things that AI in the US
is still clearly in charge of and front
and leading the pack. In terms of raw
intelligence, the frontier models and
the amount of money spent on
infrastructure, we're definitely
winning. But global opinion is shifting
the AI race away from the United States
and towards China. So a poll by Public
First asked people in 15 other countries
which nation they think is leading in AI
capability and innovation. And in 11 of
those countries, respondents said that
China has passed the United States. And
what were the countries that still
thought the United States was leading?
Well, one was US, the United States. So
we think we're doing great. Also
Vietnam, India both also agree with us
and same with Japan. But even inside the
US, confidence was not strong. Just 51%
of Americans said their own country is
leading. So three, you know, just 51% is
like barely over half of us. Canada,
Mexico, France, United Kingdom, they're
all thinking that China is beating us.
It also reflected a certain frustration
with American AI models, which the
article describes as corporate,
extractive, and hostile towards workers.
These countries saw China as a country
that presented itself as having stronger
competition and more people centered
regulation. The main claim is not that
like China is ahead. It's that the
world's trust is moving away from the
United States and towards China and that
creates a legitimacy problem for Silicon
Valley. What do you think? Are you out
of the 51% do you lean towards thinking
like American AI is I don't know more
people centric or it's going to be
better less extractive? I mean,
obviously it's not going to be like less
corporate or less extractive. We're the
United States, but is it still still
winning or in the long run, do they have
a better sort of pipeline built? I don't
know. I'll make this one quick, but just
so you guys know, Chad GBT is now under
50% for the first time. So, what's going
on here? I want to dive into it a little
bit. I'm also going to touch on the
cursor and the codeex and like what
companies are doing as far as their
positioning each other and uh what
exactly was $60 billion like just that
is that that's so overvalued in a nonAI
world that is so almost undervalued in
the long-term thinking of an AI frontier
model play. So, it's just the whole you
if you buy the story that like AI is the
invention of intelligence and whoever
gets there first gets like so much
reward then you know what that that
makes sense and honestly like probably
no matter what you buy at this point if
you truly believe that's the story you
could justify it. I mean look
realistically you have to even kind of
wonder if SpaceX what's it trading at
like two trillion market cap and then uh
Tesla's another 1.3 1.5 trillion. So if
you take all of that Elon's in chart and
plus like Neuralink and all these other
things, he's got to be coming up on $4
trillion worth of buying power and a
trillion dollars personal wealth,
you know, and if OpenAI just goes
public, if Anthropic goes up public at a
trillion dollars, like maybe he could
buy both of those companies or one of
them and just merge them in or
something. So, you know, I don't know
who's big enough really. I feel like
Google and maybe Microsoft are too big
to be acquired. maybe meta, but there
might be we might be coming down to like
in the long run and might not even be
that long. Like I'm thinking just two,
three years, we might see some crazy
consolidations. Like there could be a
true merger between like Meta and
Microsoft or something because there's
something happening with like Google and
SpaceX. I think Google owned a bunch of
SpaceX stocks, so there's they're like
tied into that a little closer. But uh
let's look at this article. Michael
Spencer, Jeff uh Mohouse wrote, "Codeex,
Cursor, and the great beyond. Open AI
will be put under pressure by Meta AI,
SpaceX, and China, not just Anthropic's
Enterprise AI momentum." So, there were
less than 60 people that owned the stock
for Cursor. So, $60 billion made them
all billionaires, which is like
mind-blowing to me because it's a fork
of VS Code. It was like an open- source
software, and it was awesome. and it was
a way to just plug large language models
in and code. But, you know, they built
an amazing tool for enterprise and it
just has so many people using it and it
has such an interesting information loop
and it's so valuable to an AI company
like SpaceX that's behind anthropic in
terms of coding. And if you believe
coding is the flywheel effect where if
you get code right, then you can catch
up on everything else. Like there's no
need to have a video model right now or
an image model. just like just code and
then we'll code everything that we need
in the future and then it'll code some
super advanced model that does even
better video and audio or whatever all
these side quests that you want to
tackle. Like it will just code them all
up for you and you will win them all.
But it puts OpenAI, which now is under
50% market share, in a strange position.
Like they seemed untouchable a couple
years ago, but they're under pressure
from every side. And now SOAR has been
cancelled and um Stargate, it doesn't
feel like they're going to get as much
money as they thought and you know
there's all this other competition going
on. So Ched GBT, no doubt about it,
still has a huge reach. But I mean I
don't find myself in it as much as I
ever used to. I mean first off that
whole ad thing was bothering me. But
also, when you're in Chrome and there's
Gemini right there and you're kind of in
a Google workspace or you're in a
Microsoft 360 workspace, they have their
own models where you're going to go to.
When you have questions inside meta,
they have their own unique data source.
And chat GBT still has its spot, but it
just I don't know, it does feel kind of
like I I got this sense that Google's in
the best position in the long run. that
that's my main bet. But I'm also I
wouldn't be surprised especially with
SpaceX having like essentially infinite
money and having having so much extra
compute. I mean they Elon built so many
servers so fast and they really are
valuable. Like they're I think he's paid
off all of them just with the money he's
getting from Anthropic. So that was like
an insanely profitable decision to just
push those things out. then China as a
whole can mount a pretty good, you know,
a little bit of that communism there can
like really just force companies to work
together and try to take on individual
companies in the US which might force
individual companies to go to these like
crazy mergers. But anyways, ChadB still
has huge reach but the article says that
its market share is failing because
Gemini, Claude, Meta AI, Space XAI and
Chinese models are putting that pressure
on them from every direction. Really,
Meta should have made that. If
Zuckerberg was serious, he's putting a
lot of money on talent. Like, they
probably should have bought Cursor.
Maybe Microsoft should have bought
Cursor. I mean, they already have
GitHub, which is pretty powerful. So,
maybe I understand why Microsoft didn't.
But Curser already had fast revenue
growth, strong use among software
engineers, and a product people trusted.
And that makes it a direct threat to
OpenAI's Codeex plans. Codeex used to be
simple. It was OpenAI's coding
assistant. You gave it a software task
and it helped turn that idea into code.
But now Codeex is expanding beyond
programming. AI wants it to handle
broader work across tools like Gmail,
Calendar, PowerPoint, and computer use.
And don't get me wrong, this is probably
the right move for them. I just don't
think they're going to pull it off. I
just think that's hard. But if I was if
I was Sam Alman, I also actually would
be doing this cuz you don't want to
think small at this moment. Maybe maybe
they'll have something that kind of
carries them through the day. But the
risk is execution. If OpenAI can merge
chat GPT and codeex really seamlessly,
if they can somehow make the average
person just sit down with the tool and
start kind of accidentally dreaming up
things that are coded into reality and
hosted and profitable. And it really is
like the Apple moment for them. They
might they might be able to continue
this route. If not, I think I think the
walls are closing in on them.
I mean, don't get me wrong, everyone
there is going to end up being a
billionaire and like Chad GPD won't just
be gone, but it's not Sam Alman's moment
to potentially outdo Elon will be gone.
All right, so if you enjoyed this video,
please share it with a friend. You're on
YouTube, there's a little share button
right there. Please send it to somebody.
But if not, leaving a comment, hitting
that hype button if you guys remember
that, thumbs up, all that stuff. Like
any interaction with the the video
helps. And just the fact that you're
still here watching it means a lot to
me. So, thanks and I will see you in the
next video.