[CC may contain inaccuracies] Thank you.
Congratulations. I mean, there was big news this week,
billion valuation for a company that you started a months ago called 018.
But in Chinese, it's even a more beautiful name.
Well, it means Lingua Wu, meaning zero in one.
The binary digits will bring about a genesis.
And that's on Avaya. Okay, so what's the limit at the moment?
It's been eight months. There are challenges which we'll talk
about. There are opportunities, but what kind
of company do you want it to become in five years and then ten years?
We want it to be a global company. That's why we open source our best model
to date, and that's what we announced this week.
It's a 34 B model that is the highest performance open source model for a
large language model exceeding Matthias Lama, too, by a significant margin.
We believe open source will bring the world together.
We think research is always about building on the shoulders of standing on
the shoulders of giants and we want to contribute from which we talk.
Is it an advantage or disadvantage having the China tilt right now?
Oh, we have a lot of people who are American trained, a lot more Chinese
trained. We have a single goal, which is to make
the best large language model and bring higher productivity and GDP to the whole
world. Have you apply for clearance from
China's regulators to roll this out publicly yet?
And what's the timetable for that? We are just announcing the open source
right now and it's available globally. Okay.
You know, obviously, there's a lot of talk about this whole, I guess, race
between the U.S. and China here.
How do you see this all playing out overall and what really is at stake
right now? I think the likelihood, given where
things are, which is regrettable, is we have two parallel universes.
American companies will supply their products and technologies to the US and
other countries that want to use American products.
And Chinese companies will build Chinese products for China and some countries
that will use Chinese products. So the reality is that American
companies and Chinese companies are probably not going to compete all that
much in the same marketplace, but they'll probably have to compete on chip
makers. Right?
So the US administration has said that some of the higher tech chips cannot be
sold to Chinese companies. What does it mean for for your company?
Right. We we saw this coming, so we actually
stockpile a lot of Nvidia chips. So we're in good shape for about a year
and a half. And I think the jury's out whether in a
year and a half China can make it equivalently or nearly as good chips.
Time will tell. You're not just targeting China, but
also global markets. And just how do you view the whole
regulatory landscape right now when it comes to Asia?
I think there clearly needs to be regulation because I can be used in ways
that are harmful. And I also think different countries
will have somewhat different regulations, but hopefully there's
enough commonality in them. So regulation is necessary.
But also I think we should it shouldn't expect regulation to a along to be
enough because a lot of these issues are hallucination, misinformation, harm can
and should be measured by technology. And we should let the technologists find
a way to contain the technology, but not completely rely on them either.
So it's really technology to provide the way out rail and detection and removal
of harmful and illegal content, but working side by side with regulations.
At what point, I mean, do you think that it could actually disrupt innovation in
some way? I mean, are you concerned that the
regulations could actually hurt innovation?
I think excessive regulation will hurt innovation.
And I think we as a technology community are willing to take some hits and be
slowed down somewhat for the good of humanity, but not so much.
So that will completely stifle innovation.
But how difficult is it and this is one of the conversations that we try and
have with policymakers. First of all, policymakers, probably
regulators don't understand how quickly or how slowly this comes compared to
someone like you and your expertise. So what's the biggest mistake that they
could do? Well, I think a moratorium would be a
huge mistake. It's not enforceable and it's not going
to be effective. You have to let technology move forward.
I think things that they should do are to promote letting technologists work on
fixing hallucination and other problems. And I think they should use existing
laws to punish those who misbehave, for example, for spreading misinformation.
We have laws in every country on that already.
Just use them, maybe make sure the penalties are a sufficient way to stop
people. From doing bad things.
And I think coming up with brand new laws is nice to think about it.
At some point that will be needed. But I don't have a lot of confidence
that any government can come up with a brand new set of laws.
I think the existing laws are plenty good for, you know, anti-terrorism and
harm and misinformation and false advertising.
I think we should use those first. At the end of the day, who will use your
products? Actually, is it.
Is it consumers? Is it going to be a consumer to consumer
kind of product? Well, our products will be platform and
application. I very much believe this narrative, AI
is not just about who can build a better model.
We'd like to build the best model. We'll try to build the best model by the
end of the day. It's really about building the right
platform and the set of tools that will allow a proliferation of new
applications built on. I just like mobile Internet and Android
and iOS allowed mobile apps to be built and PC and Windows allow PC apps to be
built. So it's an ecosystem play, and the best
business opportunity is to build the model and the platform and the
applications. And we have the ambition to do all
three. We've seen, I mean, in the earlier part
of the year, I think in China and especially right, if you if you had a
company that touched anything, the air mentioned anything about the stock would
just go nuts. That sort of a frenzy, though, seems to
have died down now. What is investment sentiment looking
like now? Well, I think globally, including China,
is going through the typical Gartner hype curve.
Right at the beginning of the hype, things go up through the roof and
everything is so great, valuations are so high and now people are a little bit
into the disillusionment. Sam, are the tragic beating numbers
down? Are EU numbers down and investments more
difficult pieces are being more cautious.
They think there can be only a few players, not enough great apps.
It's very typical, but I think we're going to very quickly see a rebound
because the value that is being created by generative AI is indisputable.
We need to give a little bit of time for popular apps that create a lot of value
or save a lot of money to be developed and for their value to be demonstrated.
And once that is, the Gartner hype curve will show that we will rebound back and
eventually be even higher than the original hype, because that's how big
this revolution is. So what excites you the most then, about
A.I. right now and what's overhyped?
I'm most excited about the opportunity to improve improved productivity for all
of humanity. I'm incredibly excited by the fact that
I can now pass the training tests.
And of course, that's a double edged sword, and I'm incredibly excited that
intelligent assistant agent interfaces are on the horizon where we can give
them the specific goal we want to accomplish and that A.I.
will figure it out. We're not being replaced by
our jobs are anchors are staying. A year ago, I would assume is I would
have said very safe forever. No, no, no.
I would say safe for a couple of years for most anchors, for the two of you.
Our soldiers say, Oh, it's okay, you can come back.
Yeah, go for it and get.
