Sam Altman downplays the dangers of AI

Sam Altman is the cofounder and CEO of OpenAI. His company's ChatGPT has changed how the world sees artificial intelligence and raised fresh questions about the impact of artificial general intelligence -- a machine that can do anything the human brain can do -- on society. (The New York Times has sued OpenAI and its partner, Microsoft, claiming copyright infringement of news content related to AI systems. The two companies have denied the suit's claims.) This excerpt from his conversation at the DealBook Summit this month has been edited and condensed.

Andrew Ross Sorkin: On the artificial general intelligence piece -- it sounds like you might be getting close.


Sam Altman: My guess is we will hit AGI sooner than most people in the world think, and it will matter much less. And a lot of the safety concerns that we and others expressed actually don't come at the AGI moment. It's, like, AGI can get built, the world goes on mostly the same way, the economy moves faster, things grow faster. But then there is a long continuation from what we call "AGI" to what we call "super-intelligence."

Sorkin: That's when we should get worried?

Altman: Well, even at the AGI moment I think there are things to be nervous about. I expect the economic disruption to take a little longer than people think, because there's a lot of inertia in society, but then to be more intense than people think. So the first couple of years maybe not that much changes. And then maybe a lot changes in the economy. I'm not a believer in the no-work (fear) -- I think we'll always find things to do. But with every major technological revolution there's a lot of job turnover. But I would bet we will have never seen it go this fast. So I think there's things like that to worry about in the relatively near-term.

Sorkin: And do you have any faith that the government or somebody is going to figure out how to avoid that?

Altman: I have faith that researchers will figure out how to avoid that. Let me caveat that a little bit. I think there's a set of technical problems that the smartest people in the world are going to work on. And, you know, I'm a little bit too optimistic by nature, but I assume that they're going to figure that out. I think there's a lot of work in front of us, but I also think we have this magic -- not magic -- we have this incredible piece of science called "deep learning" that can help us solve these very hard problems.

The societal issues around not just AGI, not the thing that we think will come in a few years that can do a bunch of jobs and create a lot of economic value, but true super-intelligence -- the system that is not just smarter than you or me but smarter than all of us put together, unbelievable capability -- even if we can make that technically safe, which I assume we'll figure out, we are going to have to have some faith in our governments. There are going to have to be some policy issues around that. There is going to have to be global coordination that -- I assume we'll rise to the occasion, but seems challenging.

Sorkin: There's been trust questions. Just speak to it so we understand -- when people say that there's not enough focus on safety, you think what? What is not happening that is supposed to be happening?

Altman: Well, I'd point to our track record. I would say, you know, we talked about we've got this big product out there. But this is a very new technology moving very quickly. It started off as something that we did not know how we were going to align very well. And it is now generally considered by most of society to be acceptably safe and acceptably robust. Safety, that's always a contract negotiated by a bunch of stakeholders. It's maybe hard to exactly define what it means for ChatGPT to be safe...

Sorkin: But the reason I ask is when we hear different people who have left the organisation and they've gone public, they take to Twitter and they say, "These folks are not focused on safety" -- is that about resources, is that about processing power, is that about your attention?

Altman: There are definitely people who think that ChatGPT is not sufficiently safe, that it allows things that it shouldn't, or doesn't do things that it should, or there are people who will push on parts of that. There are people who will say, you know, "Fine, ChatGPT. Maybe that's a little bit safer than we thought it was going to be, but what's the plan for the next system? What are we going to do about that?" And also there are people who would say, "Putting out a system like this at all is unsafe because OpenAI accelerated a race in the world, and that gives us less time to work on safety."

And so we believe -- and this is an opinionated stance -- that this idea of iterative deployment is really important. We've got to put these systems out into the world. Society and the technology have to coevolve. You have to start while the stakes are lower. You have to understand how people are going to use this, and what it doesn't work for, and what it does. There are other people who say, you know, "Sure there's some benefits there, but it's not worth the costs."