Happy Mind: The Philosophy and Science of Happiness

 

(This is the longer version of an interview originally published in Rocks Magazine, February 2007 issue.)

 

What is Artificial Intelligence (AI) and what is its purpose?

AI is the study of intelligence in humans, animals, and machines and the attempt to engineer intelligence into any type of artifact.

One purpose of AI is to understand what intelligence is and how it works. The second is to build useful technology. As a matter of history, AI has been very successful at both.

 

Can you tell me about your research in particular and how you became interested in AI?

As a teenager I liked to rebuild engines and read moral philosophy so AI was probably the only place where I wouldn't be off-topic half the time. Seriously though, now I call myself a 'machine ethicist'  which makes more sense these days than it did when I started out. I'm interested in the cross-over between ethics and technology. This includes the ethical behaviour of technologists and the design of technology which is socially beneficial. In the last couple of years there has been a certain amount of recognition that this is an important area - particularly in AI.

Is there anything which you think a silicon-based intelligence could never do? Do you think, for example, that machines could be capable of consciousness and emotions?

The best answer is that, right now, nobody knows. Quite a few AI researchers definitely believe that machine consciousness and machine emotions are realistic goals and some even predict them as only a decade or two away. Such predictions are based on their faith and optimism rather than on anything actually working at present, but that doesn't mean that it can't be done. Personally, I have doubts about the ethics of deliberately trying to build conscious or emotional machines. I also think we can build useful robots without worrying about them needing either consciousness or emotions.

We think of activities like playing chess and solving mathematic puzzles as being 'clever', whereas we don't give ourselves much credit for being able to play Frisbee, but is there a sense in which playing Frisbee is a greater achievement?

Well, for humans, I'd rather say playing Frisbee is an equal achievement to playing chess. What's interesting in AI is that we have long ago solved the problem of playing chess, but are not yet able get a robot to play Frisbee. That fact certainly changes your view of how great an achievement playing Frisbee actually is. I don't actually know of anybody working on a Frisbee robot, but  plenty of AI scientists and engineers are working on problems very like playing Frisbee. They have to solve a collection of very difficult problems and probably they'd say they were harder than chess-playing.

How far away do you think we are from interacting with artificial intelligences we could realistically come to regard as friends, or even fall in love with?

Well, the flip answer is "It all depends how sad you are". Being serious, though, robot companions are a booming research area. In Japan, for example, the public generally have positive attitudes towards robots and a number of researchers are working towards robots that will care for Japan's growing elderly population. Whether someone would fall in love with any foreseeable robot companion is a question about human psychology. Some people developing the technology would like it to happen and may build in tricks to try and make it happen - for not very respectable reasons. My latest paper starts with a slight misquote of Jane Austen: "It is a truth universally acknowledged that a single man in possession of a good fortune must be in want of a robot companion."

I have this vision of one day hearing a Karaoke version of Stand By Your Man that starts with the line "Sometimes it's hard to be a robot"... but when I do I won't assume that the robot is actually
feeling anything as opposed to its designers playing tricks with my feelings.

What do you think of popular culture depictions of AI, like Star Trek's Data, HAL9000 and the Terminator?

Literary theory suggests that what we want from fiction (including the best science fiction) is a sort of mirror where we can examine our own emotions and motivations. All the AIs you mention are very, very human-like in their (artificial) motivations and psychological processes. Yes, even the  Terminator whose final body count was less than certain real humans I could mention. This makes for good quality entertainment, but I have to say that real AI is a good deal less human-like in its working.

In your opinion, what have the greatest successes and disappointments of AI so far?

AI has provided so much useful technology that people just don't notice. From search engines to mobile phone routing, a lot of AI is quietly working to make people's lives  better today. There have been a lot of disappointments though. It's usually turned out to be much harder than anybody expected. We can't yet build robots with the intelligence of, say, ants. As for the popular culture AI you asked about in the last question, it's simply not in prospect at all.

 

© Katherine Power 2007. Published in Rocks Magazine, February 2007.

 

Katherine E. Power is a freelance writer and the editor of Happy Mind. Her website can be found at: www.katherinepower.com.