Future Current

SIAI Interview Series – Christine Peterson

The following transcript of the SIAI Interview with Christine Peterson has not been approved by the author. Video and audio can be found online at the Singularity Institute website.

The Connection Between Artificial Intelligence and Nanotech

I’m Christine Peterson. I’m Vice President of Foresight Nanotech Institute, and I serve on the advisory board for the Singularity Institute for Artificial Intelligence. My role for the Singularity Institute in my advisory capacity is to help them understand two key points. One is: what will be the connection between the development of artificial intelligence and the development of nanotechnology. And the other is a broader question, which is, How do you educate people about the longer term technological future? Nanotechnology is about the future of the material world. Artificial intelligence is the future of the information world.

The purpose of the Foresight Nanotech Institute is to maximize the benefits and minimize the potential downsides of nanotechnology. In particular, the longer term type of nanotechnology, when we will be able to build large scale physical objects with atomic precision under programmable control. It’s a challenge, both for Foresight Nanotech Institute and the Singularity Institute, to look into the future of technology. Of course, we can’t project the future accurately. It’s never been done; it probably never will be done. But there are some trends that you can see very clearly. Most people in technology are aware of, for example, Moore’s Law, which projects the future of computing. That is critical both for nanotechnology and artificial intelligence. So, even though we cannot put an exact date on things, we can see about how things are going, and we can make some educated guesses for how long some of these very dramatic developments are going to take to arrive.

Not everyone is either able or willing to think about the longer term future and what possibilities are coming. In particular, what possibilities look pretty definite. At this point I would say those pretty definite possibilities include both nanotechnology and artificial intelligence. How long will it take? We’re not sure. But it’s important that some foresighted fraction of the population take an interest in these vital issues and try to steer technology toward the more beneficial and safe outcomes, and away from the negative outcomes that, for example, could have very serious impacts on the economy, on peace, and on our freedoms.

Foresight Nanotech Institute, we have founded a project called the Technology Roadmap for Productive Nanosystems. And this roadmap lays out the technical steps on the pathway to the very advanced nanotechnology manufacturing capabilities that we think merit very serious attention. Both in terms of their technological possibilities and potential risks, as well. That is a central project for us, and it is how we think about how do you guide nanotechnology. There are big similarities between the project of making beneficial nanotechnology and the project of making beneficial artificial intelligence. Both of these are long-term, very high-tech projects that involve an interaction between technologists and policy maker, and so there are synergies there. Having spent a lot of time thinking about these issues for nanotechnology, I am able to put forward some ideas for the Singularity Institute to use in their efforts in artificial intelligence.

A critical step in the project of guiding artificial intelligence is the creation of a community of like minded people who care about this goal and who bring important skill sets to it. What we have found at Foresight in nanotechnology is something that the Singularity Institute is doing now, which is to have a series of high-profile conferences on the topic that bring together the thought leaders and start to form that active community, working together to identify what are the issues, what do you do about them, who is going to take action, where is the money coming from, for these important issues. So, I’m a big supporter of the Singularity Summit and the other technical meetings that the Singularity Institute will be having.

There are a lot of definitions of the Singularity. But in terms of a practical one, a useful one, that I like to use, I think of it as the coming time when the technological change increases to the point where we are really scrambling, and feeling perhaps overwhelmed. So, obviously, if there is a way to prepare for this, it behooves us to do it now, and that is part of the goal of both Foresight and the Singularity Institute. The task of guiding artificial intelligence to beneficial outcomes is an extraordinarily difficult and complex task. It’s going to take a great deal of thought, and a great deal of preparation by a lot of people working together, and it’s going to take quite a bit of time. We don’t know when the results of this effort are going to be needed. It could be sooner than expected. It would be a very good idea to get started at a fairly brisk pace right now on this, because it may turn out that we need the results sooner than we thought. So I’m a big fan of the idea of getting started with a strong effort now.

In terms of how much time we have before Strong AI arrives, there are a number of different estimates. I divide them roughly into two camps. One is the folks who agree with Ray Kurzweil. He, in making his estimate, looks at how hard will it be to model the human brain, and how long that will take. And he comes up with a particular estimate of a few decades out. I think this may be longer perhaps than it actually will take. One could imagine shortcuts. There is no reason to argue that this is the only way to do it. After all, that’s not how our software works today, and we do some pretty interesting things. So, I would make the case that it may be shorter. It may be significantly shorter. In fact, it may be very short. We don’t know how long it is going to take, where it is going to happen. It could happen tomorrow. It could happen in the U.S. It could happen in China, or Japan, Russia, or Israel. We don’t know. I would like to see anyone prove that this isn’t going to happen tomorrow.

We have been trying to make Strong AI for decades, and we are all very aware, those of us who have looked at the issue, that there were some misconceptions early in the field. People thought that it was easier than it was going to be, back in the ’60s and ’70s. People have let that cloud their judgment, I think, and just override any consideration of the idea that there could be a near-term result. I think that’s dangerous. I think we need to keep our minds open. Just because some people were wrong back in the ’60s and ’70s doesn’t necessarily mean that a particular technological achievement can’t be done in the year 2007 or 2010. I’m not persuaded that we need advanced nanotechnology to create Strong artificial intelligence. That’s not clear to me at all. However, what is clear, is that when we do achieve very powerful nanotechnology, the ability to build with atomic precision, that will definitely help the task of building highly advanced artificial intelligence, if necessary by creating physical models of the human brain.

Both in the field of nanotechnology and in the field of artificial intelligence the question of how many people are working on these goals and how broad is the base of technological skill being applied to these problems, can be answered in two ways. One is how many individuals are attempting to leapfrog to find a shortcut to the very advanced version of Strong AI. That number is relatively small. It’s very hard to come up with an exact number of how many people are attempting to drive directly to advanced AI, and to advanced nanotechnology. Because these are not U.S. based efforts. There are very competent technical people in both of these fields all over the world. And not all of these efforts are necessarily public. In the field of AI, where there are huge commercial advantages to getting to market first, I think there are numerous projects, which are not advertising themselves. And for very good reasons. Not just in the U.S. but in other countries as well. What is easier to do is to look at, in both fields, the number of people who are building the groundwork, the base knowledge, and advancing the fundamentals of both fields, which will, without question, in my view, achieve these advanced goals at some point. They are not leapfrogging or shortcutting. They are taking the long, hard, slogging pathway, and spinning off many commercial projects in both cases. So, those numbers are huge, both for nanotechnology and for artificial intelligence, which in this case you can count much of software technology in that space. So these are huge, huge numbers.

I believe it is critical that these extremely powerful technological developments come to fruition first in the democracies, in places where the population has at least some control over, for example, military use. We don’t want very powerful new technologies to be arising in places where, for example, democracy does not have a strong hold. Is perhaps relatively new or even non-existent. I think there may be disagreements about which country would be safest. But I think we can certainly agree that there are some that would be worse. So, our feeling is that we are hoping to speed up development, preferentially in the more open societies. This effort to guide the development of artificial intelligence in safer and more beneficial directions is very new. We are in the early stages. We are still building a community. We are building the technical skill set of the organization. We are building a funding base. We are building the teamwork, all those things are starting to come together. And I think that this approach of collaboration and open interaction is going to encourage development in the more open societies. I believe the Singularity Institute will be clearly tracking developments around the world, but it will be the more open societies who join with the Singularity Institute in this effort. Those collaborations are going to lead to faster technical progress, I believe.

In the field of nanotechnology, there is increasing support for these longer term, more advanced goals. For example, in the recent review of the National Nanotechnology Initiative, published by the National Academies, there is a section on molecular manufacturing, recommending tighter coordination between theory and experiment in this field. The U.S. government also provided kick-off funding in that field for the International Council on Nanotechnology, which is a public/private multi-stakeholder organization attempting to steer nanotechnology in safer directions. Similarly, for artificial intelligence, in that case funding comes largely from DARPA, the Defense Advanced Research Projects Agency, which is part of the Defense Department. So, they are funding it technologically. I predict that sooner or later, just as with the International Council on Nanotechnology, the U.S. Federal Government will probably be attempting to encourage multi-stakeholder groups such as the Singularity Institute in their efforts to guide AI in safe and beneficial directions.

Given that this is a long term effort, why should someone support the Singularity Institute today? We can see that it’s important for the longer term, but I think that from a personal perspective, for me, it is important that I feel part of this collaborative community, many of them highly talented individuals, very high technical skills, and very benevolent in their outlook: steering their efforts, both their public efforts and even in some cases their private companies, in positive directions for society. And I just find that very inspirational and I certainly would invite anyone who would enjoy being part of such a collaborative community to join with us.

This entry was posted on Wednesday, September 5th, 2007 at 4:24 pm and is filed under Artificial Intelligence, Christine Peterson, Singularity Summit. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.