Greg Egan Talks Upcoming Books And The Potential Downsides Of Artificial Intelligence

Illustration for article titled Greg Egan Talks Upcoming Books And The Potential Downsides Of Artificial Intelligence

Keeping The Door has a great interview with Greg Egan, in which he talks about his next two books, and why artificial intelligence may not be as close as you think — and that may be a good thing.

The hard science-fiction mastermind's descriptions of his next two books sound pretty intriguing, especially the second one:

Zendegi is set in Iran in the very near future; the first part of the novel takes place in 2012. The ultimate focus of the story involves brain mapping and virtual reality, but the backgrounds of all the characters are entwined with the Iranian pro-democracy movement in various ways. It's due to be published in mid-2010.

Orthogonal is a novel I'm working on right now; it's set in a universe with laws of physics that are different from our own. One small change in a fundamental equation - just turning a minus sign into a plus sign - leads to some incredibly rich variations in everything from the way biology works to the relativistic effects of space travel.


He also explains why we may not be as close to creating A.I. as you'd think, and why we should tread carefully:

I can't say I'm disappointed, or surprised, that we don't have artificial intelligence yet. I've written things where conscious software is created in the near future, but it's usually in the form of direct copies of human minds, so it's more a matter of us migrating from our bodies than creating a new form of intelligence from scratch.

At the moment we're so far away from creating any kind of conscious software that it's hard to know which prospects are realistic, and which are pure fantasy. When we do finally grope our way towards some tangible results, I hope we proceed slowly and carefully, because this has the potential to lead to a lot of suffering.

The present generation of humans emerged out of hundreds of millions of years of animals tearing each other's throats out, and tens of thousands of years of people being prey to famine and disease. We might aspire to do much better than that, but creating an entirely new kind of intelligence that's happy with its own place in the world is an incredibly daunting prospect.

Interestingly, he also says he thinks brain-mapping may be the most promising new technology being developed right now.

Egan also explains why he didn't write any books for four years (he was working to help refugees in Australia), why he doesn't allow any photos of himself on the internet, and why he avoids science-fiction conventions. [Keeping The Door]


Share This Story

Get our newsletter


The flaw in his reasoning is to assume that an AI built out of computer hardware and software will evolve as slowly as biological entities have. There's no reason to assume that to be the case.

Just as an example, an AI built out of a quantum computer would evolve super-exponentially faster than humans have. #books