How Blockchains Can Make AI Safer

If you’re curious about emerging tech, it’s harder to imagine a cooler job than Allison Duettmann’s. She’s the CEO of the Foresight Institute, a think tank founded in 1986 that’s devoted to advancing science and tech that’s arguably undervalued — fields like nanotechnology, space exploration, and harnessing cryptography to build safer AI.

Allison Duettmann is speaker on the AI Stage at Consensus 2024, in Austin, Texas, May 29-31.

Duettmann lives in the world of Big Future Ideas. She’s the coauthor of “Gaming the Future: Technologies for Intelligent Voluntary Cooperation”; she’s the founder of Existential Hope (with a mission of “Thinking Big for Positive Futures”); and she’ll be joining Consensus 2024 at the inaugural AI Summit to share insights on what’s legitimately useful in the world of AI+Crypto, and what’s just hype.

In a recent Zoom interview, we cover the staggering upside of nanotechnology, dig into how crypto could help the development of AI, and we may have accidentally just birthed a billion-dollar token.

Interview has been condensed and lightly edited for clarity.Let’s start with a curveball. With some tech that seems important to you, but doesn’t get a lot of press: Nanotechnology. If we crack the code of nanotech, what’s an example of a use case? How would the world be different?Allison Duettmann: Basically, it would revolutionize the entire way in which we produce things. So, yes, individual applications are exciting. But more generally, it’s almost like we’re still producing things in this super stone-agey way. We take the resources we have and try and strap them together.

But if we could get much more precise [through nanotech] about what types of resources we use in the production process… Like, currently, we’re operating at an entirely wrong level. We’re producing so much waste in the process of producing.

Imagine a 3D-printing method, for example, but on a tiny molecular scale. So you’d have massive gains on that front. Nanotech could have really interesting implications on hardware for computing, and really interesting applications in longevity technology.

How so?

For cryonics, for example, you need almost molecular nanotechnology for the revival of patients — such as on the drug delivery level or the nanobots in your bloodstream perspective. Then there’s also innovation from an environmental perspective; carbon drawdown would be enabled through molecular nanotechnology, or waste removal in the biosphere.

Given the massive upside, why hasn’t there been more development in nanotech?

One thing that’s so interesting about nanotech is that for any one individual application, there’s probably an easier way to do it. But if we had this kind of general purpose, high throughput nanotechnology, then you could be using that for many different fields.

So there’s never one incentive to develop it for an individual actor. That’s why progress has been so slow. It’s almost like this public good that once it’s created, it can cause these revolutions across different industries. But for any individual application, there’s never a unique inventive to develop it.

That’s so interesting. I say this only mostly in jest… it sounds like a good candidate for a crypto project. Because it’s something that might take a huge crowd to capitalize, with the benefits conferred to a larger community. Is there a NANO token?!

Well…

If not, let’s start one, Allison!

Yes, now that we’ve said it, the Nano DAO.

[Both laugh.]

No, but I think in all seriousness, it fits really well. There are many people in the Foresight community that are also in crypto and care a bunch about nano. A lot of the early sci-fi nano ideas fit really well with the philosophy of crypto.

Okay let’s talk about crypto and AI. What do you see as the most promising, legitimate ways that blockchain can help with the development of safer or better AI?

I guess I would say there are three buckets. One would be the security side of things, where the Web3/crypto space has a lot of experience with failing fast if you don’t have secure systems built.

Because many of the systems in the permissionless or immutable spaces — if you don’t build them correctly — there may be a million-dollar bounty attached to them. So I think this spirit of building safety and having an almost unintentional vetting process is an important mindset, as we build up AI systems.

And the second?

The second one is on the side of, how can we leverage technologies that the crypto space is also working on, to build AI systems that we would want from a human perspective?

Can you give an example?

So, there may be more privacy-preserving applications that are possible as we draw more on the cryptography spaces. They may enable us to produce better applications, for example, for human health and in finance, leveraging cryptographic technologies. And we may be able to use cryptographic technologies to help us create better, very advanced AI systems.

And what’s the third and final bucket?

The governance side of AI. So, you can possibly use cryptographic technologies to prove that you’re following certain safety benchmarks or standards. Rather than having to share all of the information that is possibly proprietary, you can only share information that is needed to prove you’re hitting certain safety benchmarks.

Great, and I know we’ll unpack all of this more at the AI Summit. Speaking of, what are you most looking forward to at Consensus this year?

I’m really looking forward to discovering what the state of crypto and AI has really come to at this point. I come at it from the foundational research side of things or more on the kind of cryptography and security side of things. But I also know that many people have been really building hard in the Web3 and AI space. So, I’m really interested to see which of my hypotheses — or which of other people’s hypotheses — have come out. Who’s building what right now? And where are there possible synergies across different products? I’m excited to see what’s happening in that space.

You and me both. See you at Consensus.