Advertisement

When Peter Singer believes AI should safeguard itself against humans

Australian-born moral philosopher Professor Singer spoke with Yahoo News Australia about what rights self-aware AI should have.

Video transcript

- If AI is smarter than people, should it safeguard itself from interference from humans, or if it has a better moral conference than humans?

PETER SINGER: Well, if an AI is smarter than humans and also is morally better than humans-- and of course, we would have to decide just what that meant-- then it should safeguard itself against humans who are trying to prevent it doing what is morally better than what humans are doing and also work out better more likely to achieve the desirable consequences. Because, as we said, this AI is smarter than humans. So yes, I think we should-- if there really is such an AI and we really are confident about its moral values-- then think we should protect it against humans trying to turn it off.

- Just my last question on that-- and I didn't put it in the list-- is, we're very familiar with our own kind of intelligence, but should we be building other types of intelligence into AI non-human intelligence?

PETER SINGER: Well, I think we would need to decide what intelligence actually is and whether a non-human intelligence is really intelligence. I think we're really familiar with human intelligence, and we're familiar with non-human animals who are intelligent in varying degrees. But none of them is more intelligent than us.

And we're familiar with differences in intelligence among humans. Some of them very good at solving complex mathematical problems, for example. Others are not so good. Some of them are good at discussing philosophical arguments. Others are not so good. Some have good emotional intelligence. We call it a perception of what's happening with other people and how to respond to that.

So we're familiar with various kinds of intelligence that humans exhibit. But if we said, is there a non-human intelligence? What exactly is it and in what sense is it an intelligence, other than simply perhaps taking to a higher degree some of the things that humans are already doing?