Economic Questions of the 21st Century
Posted by Jeriaska on September 18th, 2007
SIAI Interview Series – Peter Thiel
The following transcript of the SIAI Interview with Peter Thiel has not been approved by the author. Video and audio can be found online at the Singularity Institute website.
“We don’t know what the year 2100 is going to be like. And you can’t figure it out by extrapolating on current trends.”
“From my perspective, the key question is always, ‘What’s the amount of leverage you get as an investor, and where can a small amount make a very big difference?’ I think this is a very leveraged form of philanthropy.”
Economic Questions of the 21st Century
Patrick Wolf was the two-time U.S. chess champion and was a second to Anand in his game against Kasparov in ’96, which was a year before the computer victory in ’97 by Deep Blue. In the early ’90s, that was an incredibly significant achievement, to be a U.S. chess champion. And it is much less so today. Just to make an observation, it’s a cultural fact that chess is less significant. It is not like it was in the 1930′s when Stalin said that he would demonstrate the superiority of the Soviet system over the Western system by its ability to produce better chess players, and that this was the measure you would use. We don’t do that anymore. And I think, in a strange way, we’re at least partially in denial about the potential of AI in the 21st century. If you look at the history of the mid-to-late 20th century, chess was seen as almost synonymous with human intelligence, and now it is seen as this idiosyncratic thing. It’s like if you could do square roots in your head very quickly. It doesn’t really tell us whether you are a smart person.
It’s clear the term means a lot of different things. It’s one of these terms that has been bandied about a great deal and has been misused in a lot of contexts. It certainly has been predicted. In the 1980′s people were saying it was right around the corner. It certainly has taken longer and there have been many disappointments along the way to generalized AI. But it’s clear that there are a massive set of issues that are happening, and people who don’t think that there is something important going on are just living in a delusional fantasy world. If you are starting a career today in the United States, I think one of the most important things you can ask yourself is, “How are you going to have a career where you are not directly competing with computers? People are focused on competing with China and Mexico, but these are not going to be the important economic questions in the 21st century. As the 21st century develops it will be much more important to answer, “How do you compete or work with computers, and what are the right terms of trade?” What is a Ricardo equivalent version in which there’s a good exchange of value between humans and computers where we benefit from the relationship rather than just being hurt. And that’s the kind of question one should be asking. I think people who are not asking these questions are likely to make very big mistakes in their individual lives, and then as a society we are likely to make very big mistakes if we don’t take these questions seriously.
I think that it’s very hard to know whether we are early or late in this process, or AI never happens for some strange reason we don’t understand yet. Maybe it is just around the corner, maybe it’s happening right now, maybe it’s still 50 or 100 years away, or even further. It’s very hard to know for sure. And, again, if you look at various things like computer chess, which people said originally would never go anywhere, and they you saw this relentless progress until they got definitively better than human beings, maybe AI progresses that way, where it is clearly measurable every step of the way. Or maybe it progresses in a way that is more exponential, or even binary, and happens almost overnight. We don’t know for sure. Now, that being said, I think the critical things for us to be doing are raising awareness for the public about these issues, making people aware that we are in the middle of the greatest technological revolution in the history of the world. While people are aware of a lot of the day-to-day repercussions, which they read about in the newspaper (or now, on the internet) they are not necessarily aware of some of these long-term repercussions. And I believe that’s precisely the sort of thing that a non-profit organization should do. It’s not something a for-profit organization can do because the time horizon is too indeterminate. If we knew this was going to happen in the next two to three years, you might be able to turn it into an investable business. You could possibly get venture capitalists to invest in it. But the kind of stuff where I think philanthropy is very underfunded and is very important is tackling long-term basic issues where the benefits are very diffuse, they accrue to all of humanity, but you cannot monetize them in a very concentrated way. These are not the sorts of things corporations will do, and they are not the sort of thing governments will do, because both corporations and governments tend to be too short-term in their thinking.
I am an entrepreneur and an investor involved in a variety of ventures in Silicon Valley. I was the co-founder and CEO of PayPal, which I started in 1998, and built it into an enormous company which eBay bought in 2002. And over the last number of years I have been mentoring and investing in various entrepreneurs who are starting the next wave of technology companies. It seemed to me that it made sense for my philanthropic work to be complementary to what I am interested in on the business side, which is this next wave of technology that is going to be transforming the world in the decades ahead. Looking around for where these questions were being tackled, I concluded that SIAI was really engaged with them in a very direct way. The people involved seemed to be the most talented. This is one of the criteria I have used for investing in start-up companies, and it seemed to be the same rules should apply to non-profits: a combination of very talented people with the right problem space that they are going after.
There are a lot of details that I am not interested in dictating. I am interested in facilitating a forum in which there can be an intelligent exchange of ideas and in which there can be substantive research on how to bring about a world in which AI will be friendly to humans, rather than hostile. I think that in some ways we don’t have the answers, and the important thing for use to be doing is setting aside the framework for asking the questions and pursuing a broad research program in these areas. The kind of place where I can be helpful is in working with Tyler Emerson and some of the other people at SIAI on how to build an organization, how to scale it, brainstorming various kinds of things that carry over very naturally. But I think the main thing I’m really interested in is creating a forum for the smartest people in the world to think about what I believe is one of the most important problems of the 21st century. A lot of different people should care. I think government, corporate, and individual actors should care. I do think that for a lot of structural reasons it is difficult to get large, somewhat risk-averse organizations to invest in this kind of research. It is cutting edge. Some people would view it as beyond the pale of respectability. From my experience, it is only by pushing things beyond the pale of respectability that you get things done and move the dial, and that you really engage with questions that are interesting and you can have reasonable breakthroughs. But it is, I think, very hard for those to get funded by corporations or by governments where there is often a completely different agenda at work. It doesn’t mean we shouldn’t try. It doesn’t mean that we will fail, but, at least in my experience, getting individual donors to step up to the plate remains the best option in the short term.
The Singularity is this shorthand description for this period of incredibly accelerated change that is going to happen. In physics, a singularity is a black hole, a rip in the fabric of space-time, where the conventions of physics no longer apply. It’s very hard to know exactly what happens. In a similar way, it is a claim about how rapidly certain vectors of technology are going to change in the decades ahead that we cannot extrapolate beyond. We don’t know what the year 2100 is going to be like. And you can’t figure it out by extrapolating on current trends. If you extrapolate the trends, you get a certain picture of how the world will be in 2100. The majority of the people living in the U.S. will be ethnic minorities and the majority of the people who are white will be Mormon. That’s what you get if you extrapolate straightforward demographic trends, for example. There are other trends you can extrapolate in different ways. And the idea of the Singularity tells you that all these extrapolations are deeply flawed, because there are certain technological vectors that will accelerate so dramatically in the decades ahead, and that have been accelerating, that this will shift things.
Where people put their individual philanthropy dollars is always something of a personal decision. I think that people should look at the Singularity Institute and see whether it is raising certain types of questions that are not being raised elsewhere. And is it providing a forum for some of the leading computer scientists, and other academics and thinkers, to come together and engage with this set of topics. I think it’s done a phenomenal job over the past few years on a shoestring budget. From my perspective, the key question is always, “What’s the amount of leverage you get as an investor, and where can a small amount make a very big difference?” I think this is a very leveraged form of philanthropy. I think that was is unique about the SIAI’s work is that the pay-off of getting it right is very big and that it is somewhat longer term, which is the place where philanthropy is very underfunded.
So, I think there are a lot of forms of scientific research that go on in university settings or in various corporate contexts where either the pay-off is very immediate, so it’s either you work on a specific project and you get the answer in six months or twelve months, or, alternately, it’s somewhat longer term but it somewhat sits within a conventional frame of what’s respectable and what should be done. I think this is just slightly outside of that. You know, we might be wrong. We might be crazy to think that work on this makes a difference. But my intuition on it is that not all the important scientific work is being done within the conventional channels. And when we look back from the perspective of 2100 on today, we will see that the really key breakthroughs came from outside the traditional channels, from things like SIAI. The question is, “Can the Singularity Institute be something like a Sante Fe Institute for the 21st Century?” with respect to the set of questions surrounding artificial intelligence and how humans and computers will coexist on this planet There’s nobody who is doing this. And one of the basic rules in starting a business and building a non-profit is you don’t try to compete with lots of other people doing the exact same thing. Nobody else in the world is doing this. And so, if we do this well, and we succeed in doing this, we will become the de facto standard for this.
I think there is a gradual evolution that we will continue on. One of the critical things in having these paradigmatic breakthroughs is to get people to think about them when they’re young, when they’re in high school or the first few years in college. So I think becoming a forum for people at that point is very important. One of the unfortunate realities is that people often don’t change their minds much after a certain point in their lives. I believe it was Max Planck the physicist who said that science advances a funeral at a time, because people get so stuck on certain ideas. While I think we should be definitely open to influencing scientists and researchers throughout the world and in all sorts of contexts, it makes particular sense to focus on people sort of towards the end of high school, early college years, and try to become a forum for a lot of the top people in computer science and related fields. We’re just starting in that.
It is a very good question to ask why this is not being focused on by more people, why it is still so much at the fringe. I think one of the conceptual challenges is that we don’t have a good intuition for what the 21st century is going to be like. If you ask people, they would, for the most part, assume it is just going to be some sort of linear extrapolation on current trends. It’s approximately correct in the short-term, but the approximation becomes catastrophically inaccurate over the medium and longer-term. I think people have this very bad intuition. Human beings are not good at dealing with exponential math, which is one of the features of the Singularity. They’re not good at dealing with probability events where it is over a probability of things that they don’t really understand any of the components of. So, I think it is the kind of thing that people don’t have a very good handle on intrinsically. I do think, if you think of it from the perspective of human psychology or human history, there obviously has been no time in the history of the world where there was the risk of human beings destroying the whole world and blowing it up. And so I think, even in the 20th century, one could argue people underestimated the risks of a nuclear war. It turns out there wasn’t one, but we’re fortunate we’re here today. I think this was something people could not quite grasp could happen.
There is nothing in our experience biologically, historically, or culturally that has prepared us for the notion that humans and the machines we build might destroy the whole world, or alternately, might radically change it for the better. So, I think that the range of outcomes is much larger. You have a probability distribution that is over a much wider range than people are used to thinking about. There are a whole set of factors like this that combine to make it very much outside of what people are thinking about. And then I think the other challenge, of course, that the Singularity Institute has is that the time horizon is not well defined. It could be ten years, it could be fifty years. Very often, in many contexts, people would like to have a well defined timeline. “You told us computers would take over the world in ten years, it was supposed to happen in 1995 and it hasn’t happened yet. So, obviously you’re completely wrong.” The correct thing might just be to understand that we are uncertain about some of these estimates. So, there’s a lot of stuff like that, which I think complicates it tremendously.
I think that people can start by educating themselves and learn about it. They can participate. The Singularity Institute has been putting on these annual summits, which I think have been a great, very compact forum over a day or so, where you get a group of people together and get some great conversations going about these topics. The way that I find I personally learn a lot is by being engaged with other people who are interested in the same set of questions and have conversations about these kinds of things. That’s how I find my learning process to work the best. But, obviously, we’re always interested in getting people’s money. But that sort of goes without saying. One thing I would add, it seems it is very appropriate that the Singularity Institute is based in Silicon Valley, right in the middle of the most likely place this technology revolution is going to happen. It is at the center of things in that sense, and it’s going to be driven by a combination of visionary entrepreneurs, brilliant computer scientists, and thoughtful researchers and academics. Somehow being at the intersection of those communities seems to me the right place to be. We might be wrong. Maybe it will happen in Paraguay, or in some island in the South Pacific, but my bet would be it’s most likely to happen in Silicon Valley.
The reason these issues are important is because there is a good chance, it’s not a certainty, but there is a good chance that in the next 20 to 50 years, they will transform the world. It would be like saying we don’t need to worry about the energy thing because we’re not going to run out of oil in the next year. But there is a good chance we will run out in the next 20 to 50 years, so we should be thinking about energy longer term. Energy is a long-term problem. It is possible to do these long-term capital investment projects, which is why it will likely partly be taken care of by the private sector. This is like energy in that it’s long-term, but it’s unlike it because we have to still do a lot of the basic research. We haven’t done that yet. We don’t know what needs to be built or done, or not done, for that matter.
With respect to the philanthropy question, I do think that one of the mistakes that gets made a great deal is this view that you sort of do well by doing good, and do well by doing good, that philanthropy should be almost like investing in where there is an immediate pay-off – like it’s business opportunity with sort of a charitable overlay. I think there are some things where that is the case. But in many cases people just end up with neither. They end up with something that is not a particularly good business and is not particularly good philanthropically, either. In my mind, at least, I find it to be intellectually more rigorous to separate and demarcate things that are for-profit and have certain parameters as businesses to invest in versus things that are philanthropic, that are to make the world a better place. In that second category, it’s a comparison between various things in that category. And this is a very important topic that is not being addressed, and I think it’s important for people to step up to the plate, in particular people in the tech community.
