Politics of AI
Posted by Jeriaska on January 23rd, 2009
Bill Hibbard, Emeritus Senior Scientist of the Space Science and Engineering Center at the University of Wisconsin-Madison, contributed a paper for the AGI-08 post-conference workshop arguing that machines significantly more intelligent than humans will require changes in legal and economic systems in order to preserve human values. An open source design for artificial intelligence could help this process by discouraging corruption, by enabling many minds to search for errors, and by encouraging political cooperation. In his presentation, he encouraged the Singularity Institute for Artificial Intelligence not to remain mute on the subject of politics and to take a proactive stance as a political organization.
The following transcript of the AGI-08 post-conference workshop presentation by Bill Hibbard has not been approved by the speaker. Video is also available.
If you want to know the details of my arguments, it is in the proceedings. In that paper, I talk about the politics of AI, which are related to open source. What I am going to talk today about is the politics of AI in perhaps a little more prosaic way than Hugo did. I think along with several of the other speakers that AI will develop pretty slowly. Very hard technical problems tend to develop very slowly. I think that the politics of AI is inevitable. None of the politicians currently running for president talk about AI. They are probably afraid of people thinking they are cooks or stirring up all kinds of religious opposition. However, I think there will be strong politics of AI before we will have AI.
For one thing, subhuman level AI will take jobs. We already have robots in factories, and a lot of jobs that people do don’t require a full level of human intelligence—they require considerably less than it. Some sort of machinery will be able to take a lot of those jobs. There will be politics from that, along with the related issue of the growing income inequality. Income inequality is a big political issue in the current presidential round, and I think is going to be a growing issue in the future. People are going to make the connection between that and intelligent machines or subhuman-level AI. There are privacy concerns—Google knows me better than my wife knows me. Moshe is an example of Google’s interest in AI. Then there is military AI. There was an interesting article in the New York Times in 2005. The title was “Who Do You Trust More: G.I. Joe or A.I. Joe?” AGI Joe is the one to watch out for. That was getting at the military ethics that we heard about the other day.
There is also financial market instability. After the 1987 crash, program trading became a political issue. As we work toward the singularity with increasingly intelligent machines, we can continue to see wilder and wilder things happening in financial markets. All of these things are going to make AI into a political issue. The real purpose of changing my topic was I knew that there were going to be at least three directors of the Singularity Institute here, so I thought that I would make a suggestions to them—a pitch. Pardon me, any of you directors, if I have mischaracterized your positions. Eliezer, I think you have been pretty clear that you don’t think much of the role of politicians meddling in the question of AI. Kurzweil, his book is nice, but it is pretty dismissive of the role of politicians in artificial intelligence. He says you cannot really control it, because the AI will figure out how to outsmart the control, and the other thing he says is that humans will become the AIs, and we will not have to worry about them, because they will have human values. Of course, history shows us that sometimes the values of some humans don’t work out so well for other humans.
Now, if you look at the website and talk to some people, the Singularity Institute does not take a position on politics. It’s sort of mute on that. Given that there is all this stuff from various directors, and I think even from you, Ben, I have suggested a certain reluctance to see politicians involved in AI questions involving ethics. To say that there is no position, given the position of so many directors, I think is a little disingenuous. The reason I am making this pitch is I have a certain admiration for the Singularity Institute. This is really the group that is out there talking about the singularity, and from the summits I understand you had something like a thousand people in the audience. There is going to be this politics of AI. It’s going to happen sometime in the 21st century. The public is going to get interested, and they are going to look around and look at what people said. If they see somebody that says, “You body politic, don’t worry your pretty little head about it. We know how to do it.”
When the politics come, you really don’t want that to be how you appear to the politicians. You don’t want that written on your tombstone. I think that the Singularity Institute would be wise to be a little proactive, because politicians are going to put their nose into this. I think that’s where you guys should be going. You should be engaging with the political process and engaging with the public. I had a discussion the other day with someone from the Republic of the Soviet Union who was pretty pessimistic about the ability of the public to have an effect. On the other hand, we have seen, in Western countries, the public having a beneficial effect on politics. There has been the consumer environmental movements, weapons control movements… these things have not worked out perfectly, but the fact that you have Ralph Nader there, a lot of good gets done. I would like to see the Singularity Institute, since you are the guys that are in a position to take this role, to become a lobby almost, among your other hats. That’s pretty much what I had to say.
December 25th, 2010 at 11:09 pm
Hope and change.