AI Convergence

aibanner.jpg

Convergence08, the technology unconference, began with a different kind of AI debate: not on whether to create AI, or which technical path will work fastest, but how we can use AI technology to build the world we want to live in. Jonas Lamis of SciVestor moderated the panel of artificial intelligence experts, which included Barney Pell of Powerset, Steve Omohundro of Self-Aware Systems,  Peter Norvig of Google and Ben Goertzel of Novamente.

The following transcript of the Convergence08 Artificial Intelligence panel has not been approved by the speakers.  Video is also available.

convergence-banner.jpg

AI Convergence

Jonas Lamis:  We have a great group of speakers here.  Barney Pell was the founder of Powerset, which was acquired by Microsoft not too long ago. He is now an evangelist, along with holding other titles at Microsoft. We have Ben Goertzel, who is the director of research for the Singularity Institute for Artificial Intelligence and also is the founder and CEO of his company Novamente. We have Steve Omohundro, who is one of the big brains behind the world of AI. Currently he runs his own company, Self-Aware Systems. Peter Norvig is director of research at Google and is another of the great brains in the computer science world.

I will be letting each of these guys tell you a little about themselves.  The way I am going to do that is by asking them a simple question, and we will start with you, Peter.  Please take no more than two minutes to answer the question: “Why are you here?”

Peter Norvig:  I am here because you all are here.  This is an interesting group to talk with and understand what is going on in the world.   I want to understand better the future and where we are going, and you guys are the ones that are doing it.

Steve Omohundro:   I think this is an amazing moment in history, where we have a convergence of these new technologies that are arising.  I think there are amazing things happening in the short-term.  Intelligent technology has the opportunity to revolutionize many of our political and economic systems, ways that we gather information and act in the world. Then there are longer-term things that are coming up that we are just dimly beginning to see, and we can foresee a lot of challenges along that path. We need to start thinking about a roadmap for deciding what kind of a future we want and figuring out a way to get there in a safe and positive way.  I am hoping that this group will contribute to those big questions.

Ben Goertzel:  “Why am I here” is a big question which you could interpret a lot of ways, but I’ll try to take the simpler interpretation.  I have been thinking about these sorts of issues that we are gathered here to discuss for quite a long time, as I imagine many of you have.  I am 41 years-old, and I guess for at least 35 of those years I’ve been thinking hard about all these things, in terms of AI, immortality, space travel and nanotechnology.  I started seriously trying to design a thinking machine when I was maybe 16 or 17 years-old.  It has been a lot of work and I think I have made a lot of progress toward those goals.

One thing that has been very interesting to see in the last five years or so is that a smaller percentage of people think I’m insane for thinking about these things.  I don’t live in San Francisco, I live in Maryland.  It’s a lot more conservative, but even there now and then you run into people who have heard what the singularity is, and if you tell them you are working on AI and on an AI that thinks like a human, they say, “Wow, good luck,” instead of “Wow, there’s the mental hospital.”

I think we are seeing a change in the attitudes of a lot of people toward this kind of work, which I think is a very important and very exciting thing.  It not only makes those of us who are working on it feel better, but it can actually accelerate progress.  Even though I am crazy enough to think that I know how to make a thinking machine, I don’t think I have every bit of the answer in terms of rolling them out in a way that benefits society.  We can all collaborate to make the singularity.  Who is going to do it is in large part the people in this room.

Barney Pell:  My thoughts would echo most of those that were said before.  I am here because I care passionately about these kinds of topics—not just artificial intelligence but in general, technology as a means for transforming the human condition.

I have big goals in my lifetime.  I would really like to see intelligent machines, I would also really like to live forever and travel in space, to be able to communicate with anyone, anywhere, at any time and in general have all kinds of superpowers enabled through technology.  I have had these kinds of feelings ever since I was a kid.

One thing that I think has just been echoed here is that we really are in a special time.  I feel really lucky that we are now at a point where technology is making some of the skeptics have to reconsider—and making some of the futurists feel empowered to unite together and make things happen.  I would like to do everything I can to connect with like-minded futurists and support this emerging community.  I look forward to all of your inputs, and when the computers look back in 25 years, these kinds of conferences are going to be seen as seminal events.

Lamis:  25 years…

Goertzel:  He’s a pessimist.

Lamis:  Barney mentioned better living through Microsoft technologies.  One of the themes that I hear a lot from the AI community is improving the human condition.  One of the questions that always comes to my mind is, do we have a growing gap between the “Haves” and “Have-Nots”? If we do, should we be looking towards our AI development activities to try and correct it?

Omohundro:  Certainly there is a growing divide between the “Haves” and “Have-Nots.” The “Have-Nots” are getting wealthier, but the “Haves” are getting even wealthier faster.  What we do about that is a sort of political question.  I think intelligent technology can help us aggregate the whole world’s opinion about that sort of question.  As nanotechnology and AI come together, we will be able to have far more productive manufacturing, which has the potential to make everyone much wealthier.  How we distribute that wealth is a perennial question.

How do we balance fairness with individual freedom?  I think that’s one of the difficult challenges that humanity as a whole is going to have to deal with as these new technologies shift what the political and individual landscape looks like.  I think the potential to eliminate poverty is huge, but doing it in a way that really reflects everyone’s human values will be one of our challenges.

Peter Norvig:  —And I think it is as much a question of values as technology.  If you have technologies that allow exponential growth, then you are going to get more divergence, because exponential curves which have a different exponent diverge faster than linear curves, which have a slightly different slope.  That means there is going to be more disparity.  The moral question is whether we are interested in equality of opportunity or in equality of outcomes.  Different societies have different takes on that.

I think the way Western society is organized, most of it is toward equality of opportunity, and I think we are on a pretty good track with that, where we can take the billions of people who are not participating in the current economy and give them the ground level of support that allows them to have that opportunity.  Then it is up to each individual, what their specific exponent is and how far off the curve they are able to go.

Lamis:  Ben, what do you think?  Equality of opportunity or outcome?

Goertzel:  The main point I would want to make is it is not us directly that are making that choice.  I’m very aware of the differences between the conditions of various people in the world.  I just arrived yesterday on a flight from Brazil, which is where my company’s main software development office is and where my wife is from.  Certainly in the outskirts of Brazilian cities you see people who have never had the opportunity to learn about any of the ideas that we are talking about here because the public education system isn’t good.  Of course, they are far better off than people in many parts of Africa.  My heart goes out to these people in these situations.

In terms of my own efforts, it seems like the greatest leverage that I can have, since no one has elected me president, and that’s not too likely to happen for various reasons, is to work in advanced AI technology.  As Steve said, as general intelligence technologies get to a certain point, the age of scarcity is going to be gone.  I don’t think any of us can hardly understand what that is going to mean, to live in a world where material scarcity is not one of the defining characteristics of life.  Not that anyone can have everything, but having a molecular nanotechnology-based synthesizer to make physical objects according to our desires is going to make a big difference.

After that point, the difference between “Haves” and “Have-Nots” will be a matter of psychology and culture more so than opportunity. I think we can get there within decades, and that is one of the most important motivations to me for working on this, to be able to uplift everyone and live more according to their choices than their circumstances.

Pell:  From my perspective on this question, I am honestly a bit torn about the extent to which the AI technologies and advances in this kind of field directly impact global living conditions, poverty and so on.   I was just at the San Jose Tech Museum’s Tech Awards for humanitarian, social entrepreneurship companies, and the things that were coming out there were like better processes to do farming and bringing light to people’s houses. I do not think that artificial intelligence will be competing with a whole slew of much earlier technologies in bringing the world out of poverty, but I do think that improvements in information technology in their time will have a totally transformative effect on people’s lives.

If you look at search, it is not necessarily changing everyone’s living conditions.  Actually, I think search benefits most of the world rather proportionately.  I do not think that it is people endowed with great resources that take advantage of search a lot better than others.  Once people get access to computing, they can access all the educational resources in the world pretty much as well as most people can.  I guess overall it is kind of like steel.  Steel helped everyone in the world.  These fundamental advances will help everyone in their right time.

Lamis:  Clearly we are all here to talk about the convergence of all these interesting technologies that we all believe are going to be accelerating in power, scope and capability over the decades ahead.  In my mind that triggers an entrance in the broader community and involvement from the political sphere in the years ahead.

We just had an amazing election that is likely to bring about some significant change in technology policy for the country.  We have heard that Barack Obama is going to potentially appoint a CTO for the United States to help advise technology policy.  Imagine if you will that each of you has been selected as a candidate for that CTO position and you are sitting down with the president.  I want to know what advice would you have on these kinds of technologies, in particular artificial intelligence.  Over the next couple of decades, how should we be thinking about U.S. technology policy?

I am going to start with Peter because if you have not been to Norvig.com, he has done an amazing level of depth of research and writings about the election. I know you have been thinking a lot about this stuff.

Norvig:  I certainly have.  That was one of my hobbies over the past several month.  As addictions go, it’s a great one because on November 4th, you are cured.  It is great to be here in this room, because we actually had on election night a “Silicon Valley for Obama” party right here in this room.

My first advice for the president, were I in that hypothetical situation, would be: “Don’t choose me, choose somebody else.”  Most of my advice would be things that I think the president is already doing.  The most important one is to believe in reality.  You’d think that would be something you would not have to tell to a president, but apparently it is.  I think Obama does, so that’s off to a good start.  I think the next thing is to invest in R&D.  His job is somewhat to lead the world, but mostly to lead the United States, and it is important to reestablish the United States as a leader there and make us a country that can step forward in terms of development.

In terms of our policies in immigration and being a destination for the world’s students to come to learn and then go off to become entrepreneurs, everyone can benefit from that.  We have slipped over the last eight years or so in terms of funding research and creating that environment, so it’s time to restore that.

Omohundro:   I see two areas.  One is using intelligent technology to make better decisions in our society, and the other is new forms of research and investigation into the future of artificial intelligence.  In the first end, our voting system is really antiquated—both at a mechanical level in how we do ballots, but also in its structure.

There is a very nice website, rangevoting.org,  where a friend of mine has analyzed a number of voting systems, and ours is by far the worst at  reflecting what people’s actual values are. I think there is huge opportunity for getting better ways of aggregating the beliefs and desires of the population. There is something I call “semantic voting,” where instead of voting for this candidate or that candidate, or voting “yes” or “no” on this issue, you should really be able to express much more nuanced opinions about something. There should be a way to aggregate that so that we can guide decision-making.

The bailout was a perfect example, where the population as a whole was hugely against it.  I and many people I know wrote to our congressman about the structure of the bailout, which I think was 99 to 1 against, as it was proposed by the Secretary of the Treasury, and yet they ended up passing it.  I think economists as a whole view that as very bad, and yet our political structure allowed it to go forward.  I think that is an example of something where we can bring greater intelligence.

I think we desperately need greater understanding of the potential pathways as we move into the future and new technologies come to the fore.  Right now it is just a smattering of little organizations, while it would be really nice to have a countrywide analysis and planning for what this future pathway looks like.

Goertzel:  To make detailed recommendations of what the president could do would obviously take longer than a few seconds.  In reality, Obama like any other leader is subject to all sorts of constraints in terms of Congress and so forth.  The basic message that I would get across if I were in that position is that we are completely insane to put a trillion dollars into corrupt banks and incompetent auto makers.

If we want to inject a large amount of government money into the economy, we should put a half trillion dollars into AI research and a half trillion dollars into life extension research, put a half trillion dollars into nanotechnology research, and don’t put it in by having government labs do it but have the government invest in private enterprise, though with a timescale that is not what typical VCs do. Let the government invest large amounts of money in companies that understand the payoff to be ten, twenty or thirty years out. Something like that could make a huge difference in the progress we have toward the singularity. I do not see Obama or anyone else thinking in those terms right now. I think that we will get there anyway, but it is too bad that there is not more of a Singularitarian vision governing the stimuli that we have decided to inject into our economy.

Pell:  It is pretty hard to advise the president.  There is a lot to say.  I agree with what Peter said about reestablishing our leadership in R&D.  I would also add education.  Future leaders are not being given the kind of education that they ought to have, and I think there are tremendous opportunities for improvements in education.  Like Steve was saying, I think we should be transforming the nature of discourse. Our government is run and our people are elected using methodologies that are two hundred years old.  With all the transformations in society, these things should change.

There is no reason why individual voices cannot be heard at all levels of government and why information cannot be shared in tremendously new ways with new levels of transparency. In terms of specific R&D, I would point to aging.  So much money is going into health care and treatment of symptoms of aging, rather than the causes.  If we could make some progress in that, it has the potential to increase our lives and reduce our costs.  The other one would probably be energy.  Nanotechnology has the ability to transform a lot.  We could invest in nanotechnology and its applications for manufacturing and miniturization, and we might be able to both generate more energy and use less of it.

Convergence: Artificial Intelligence Panel

ai-panel.png

2 thoughts on “AI Convergence

  1. Pingback: Future Current » Blog Archive » Longevity Convergence

  2. Pingback: The Singularity Institute Blog : Blog Archive : Goertzel, Omohundro, Pell, and Lamis at Convergence08

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>