AI, nuclear weapons, and the lab looking to mitigate risks

Choices we make today will inevitably impact the future. And some of those choices may not lead to such rosy outcomes.

Existential risks lie around every corner. The good news is they aren’t being totally ignored. Across the country and across the world, there are institutes studying and finding remedies for existential threats.

Scholars at the University of Chicago may have good reason to study the risk of nuclear weapons. It is, after all, where the first self-sustaining, controlled nuclear chain reaction was achieved. But they’re also studying other existential threats that risk humanity's existence, like climate change and artificial intelligence safety, all in hopes of mitigating risks that face human civilization.

“Marketplace Morning Report” host David Brancaccio spoke with Daniel Holz, director of the Existential Risks Laboratory at the University of Chicago, and professor of physics, astronomy, and astrophysics. The following is an edited transcript of their conversation.

David Brancaccio: Given your University of Chicago connection with Fermi's first sustaining atomic pile under the old stadium there, I wanted to focus on the existential threat of nuclear annihilation. Now, we'll talk about that, but AI again. AI also encroaches into the existential threat conversation, if it were to trigger nuclear annihilation. How do you think about that?

Daniel Holz: This is a pretty unique moment because of all these different threats. And so, as you said, the nuclear threat is growing over the last few years. And then AI is also, in and of itself, a threat. And then these two intersect. AI is starting to insinuate itself into all sorts of military command and control, and there's this sense that it's kind of inevitable that AI will essentially will have its finger on the button, as it were.

Brancaccio: I mean, you. Dr. Holz, also chairman of the science and security board at the Bulletin of Atomic Scientists, famed for your doomsday clock, that clock advanced in the last year.

Holz: Yep, the clock is now at 85 seconds to midnight. And we highlighted, when we announced the time, a number of things that were of particular concern, and those included the nuclear risk and AI and, of course, climate change, biological threats. But we particularly highlighted the way all of these kind of intersect and feed on each other and amplify each other, and the fact that governments just aren't engaging on these.

Brancaccio: I've done some work and have a lot of interest in interdisciplinary approaches to big questions. When you're thinking about, for instance, guardrails for AI, do you just want trained engineers at the table talking about — it sounds like the way you envision — say, a curriculum on existential risk? It would be multi-disciplinary?

Holz: Yes, that's one of the critical parts about this, and all these different expertise are relevant. I mean, the problem is kind of this all-hands-on-deck moment. So, that comes into both our work, our research, but also the classes that we teach. And the truth is, for almost all of these issues, there's a technical aspect, but to make progress, a lot of the issues are political or legal or sociological, the solutions are in a different space, and you need to have some mastery of both of those to hope to actually have impact and make a difference.