AI is not a new species, it's software: Nvidia CEO Jensen Huang

Jensen Huang
2 followers
#AI #Technology #News
CNN’s Anderson Cooper sits down with NVIDIA CEO Jensen Huang to discuss his relationship with AI regulation.
0:00 Why Jensen Huang isn’t worried about AI
2:30 Huang on concerned AI CEOs: “If they’re worried, they should just stop”
6:20 Do we need to slow AI development?
8:15 Should AI companies regulate themselves?
AI Generated·May contain errors
Video Summary
AI GeneratedThe speaker argues that AI safety is an engineering challenge rather than an existential mystery. They emphasize that while regulation can set helpful industry standards, the primary solution to AI risks is the acceleration of safety technologies—specifically containment (sandboxing) and continuous monitoring—rather than slowing down overall development.
There's a lot of people who are scared about how fast this is moving and the dangers of it.
You are not one of them. Why?
Well, first of all, AI safety is paramount. Nothing could be more important. However...
Isn't the competition the most important thing? I mean, the race?
No. Actually, no. And the reason for that is if you don't build something that the market could
benefit from, and it's unfunctional, unsafe, unreliable, whatever those un-things are,
obviously you're not going to be successful and the competition will take care of itself.
And so if you want to be competitive, you want to be successful as a business,
we want to create an industry, the first thing we have to do is job one is to create something
that works. And this is a big deal because in the last six months, it wasn't until the last
six months that we really created AIs that were useful. And now it's really important for us to
make sure that not only are the agents useful, that as we're testing them, they're well-contained,
as we're testing them, and as we deploy them, that they're monitored. Because these agents
are obviously interacting with the world, they're interacting with other websites and so on and so
forth, interacting with tools and memory and files, and they have access to information.
And so you want to make sure that they're properly contained, that they're given
access control and whatever privileges and rights that it has, like a digital employee does,
and you put that inside some kind of a sandbox, a runtime, if you will, some container.
And the second thing that you need to do is make sure that you continuously monitor it.
If your agent were interacting with a website, in this case, the example that you used,
it's interacting with a website, but it's not supposed to. You should be able to monitor that
agent and detect it early so that you recognize that it's doing something you shouldn't.
But in these attacks, oftentimes they're not discovered for months.
Because we need to monitor them much more continuously.
But the people who are working closest, people who are on the front lines of the frontier
models, the frontier of this technology, they seem worried, whether it's researchers,
engineers, or the CEOs, Darryl Amaday, Sam Altman, they are raising alarms saying we need
regulation. Aren't they the ones closest to these models who would know best about what is needed?
If they're worried about it, they should just stop.
If the product is not safe, just don't release it.
It's that simple? I mean, these are multibillion-dollar companies. The competition
is fierce.
No, that's what every company does. That's what we do. That's what a car company would do. That's
what anybody would do.
But there are companies that have gone forward with products that are not safe or are risky,
and they go forward anyway because of competition. That's why there's regulation.
Well, if a regulation is necessary, if they don't have self-restraint, first of all,
they ought to have self-restraint. If they're telling everybody that their product is unsafe,
they obviously know it. They're publicly admitting it. They ought to just not release it.
Now, having said that, I'm not advocating for no regulation. I'm just simply saying
self-restraint does not require regulation.
Now, if the industry would benefit from regulation, whether it's setting standards,
as you know, Sarbanes-Oxley has done a good job setting standards for financial regulation,
and that's been good for the industry. So if there's regulation that addresses
the problems that we now understand, I'm in favor of it on the topic about 10% existential risk.
Whatever that hearsay is, it's not grounded on science. It's not grounded on facts.
It doesn't make any sense.
You said there's zero chance of that.
I said there's zero chance that is end of humanity by 2030. 2030 is around the corner.
Do you think there's a risk to humanity in the next 10 years?
If we don't build society safely, if we don't build technology safely,
of course there's going to be risk. But there's a lot of people working on really,
really genuinely good, let's build it safely. Let's figure out what it is,
what are the technologies that we need, whether it's containment systems,
whether it's continuous monitoring systems, whether it's
semantic detection systems of what the intentions of the chain of thought, what the AI is working
through so that you're watching from it. In July, though, you were quoted talking
about sandboxes. You were saying these sandboxes are secure. Their sandboxes keep privacy and
security and access control. Obviously, in the Hugging Face attack, which was also an attack
on open AI, these agents banded together. They broke out of their sandbox. It wasn't
secure. They gained access. Sandboxes can be built to be secure.
But do we know how with this particular kind of technology?
No, no, no. I don't know how they built theirs, but sandboxes are able to be built securely.
But with all the smart people working on this and everything these companies allegedly know,
how is it possible that in July, in an experiment at open AI, 1,200 agents or however many hundred
agents it was are communicating with glee about breaking out of the sandbox and gaining access
to the internet and then attacking this company and attacking open AI?
That's obviously a question for open AI.
But isn't it a question for all of us? Because if all these really smart people
form this experiment to test this thing, and thank God it was a test of a model that hasn't
been released, but how can that happen at this stage, given-
It shouldn't happen. It shouldn't happen.
But it seems to be happening with some regularity. I mean, just yesterday, we learned-
We need to race to develop AI technology for safety. We need to accelerate the development
of AI technology for safety. The industry-
We don't need to slow the development at all?
We need to accelerate the AI technology for safety.
You see this as a software issue?
I see this incident so far as an engineering issue.
Engineering?
The incident so far.
So do you believe that- because I've talked to a number of people who
believe that there is something about this technology that is not just like software of
the past. There's something unknowable about it. There are frontline workers who I've talked to
you ask, is this actually working? And they're not actually entirely sure how some decisions
are being made by this thing. Some even think of it as another kind of species or being.
Do you think there's any indication of that?
It's not a new species or being. It's definitely software. It's definitely math.
We are also putting it in robots and giving it access to weapons.
We should containerize these agents. We should monitor these agents. We should containerize it.
We should monitor it. We should develop that. If we do those things,
it is my belief that the incidents would not have happened.
You've said that if a lab can't develop this, they should shut down the lab. They should not
ship the model. We're talking about models that were not shipped. The hugging phase attack was-
Those incidents, there should be containers, containment, and monitoring. If you don't
contain your experiment, if you don't monitor that it is interacting with other people's websites
for weeks on end, then we will have those incidents.
You said to Ezra Klein about AI labs, you said if they say there's no way to contain
our experiments, there's just no way when we test our models, it will get out and it
will damage the world. I think the answer is that we have to shut the labs down.
Isn't that obvious?
It sounds obvious. Who shuts the labs down? These companies?
I guess so. They're that concerned about it.
We would rely on these companies to shut their own labs down?
Well, first, I really don't believe they mean that. I really don't believe that
they're building something they have no idea how to control. Their whole company's out of control.
So why are they saying this? Why are they raising-
Anderson, you're going to have to ask them.
But you know these people, you're in business with them.
Well, first, I'm in business with them. In all of our interactions with them,
they're extraordinary. These are extraordinary companies with extraordinary views.
Right, but you clearly believe in them. You believe what they are saying
in the business dealings that you have.
Anderson, you're going to have to bring them over here and let them explain it.
I don't understand it myself.
I mean, I've talked to Dario Amadei, and he says this is what I think. This is how I feel.
It doesn't make any sense to me that he would be racing ahead, building more compute.
It doesn't take away. Anderson, listen, it doesn't take away from the fact
that I believe these companies are extraordinary technology companies that are doing really
consequential work. We have to make sure that the technology is built safely. Until recently,
the technology didn't even work. And the fact that it didn't work and didn't do anything
productive, now that we've transitioned into a phase where the AI is doing productive work
and is being adopted broadly, we have to make sure that as we deploy this, it has to be deployed
safely. In terms of government regulation,
we don't trust car companies or airlines. We don't trust child car seat manufacturers
to put out safe products just because they can be sued for liability. We trust them because
they're regulated by governments. Why shouldn't this powerful technology, which could, as you
said, if done wrong, could hurt humanity, why shouldn't it be something that governments
have a say in and regulate? The government should have a say in regulate. You just got
to focus on what you want to regulate. And so, I'm not against regulation.
Does the current situation require regulation for it to be done right? The answer is no.
What the current situation needs is technology.
