IEET > Vision > Sociology > Futurism > Technoprogressivism > Artificial Intelligence
A.I. Apocalypse: More Myth Than Reality
Steven Pinker   Aug 20, 2016   Big Think  

Steven Pinker believes there’s some interesting gender psychology at play when it comes to the robopocalypse. Could artificial intelligence become evil or are alpha male scientists just projecting?

Too pat. And not true. If he bothered to check on people who are actually involved in this development it will be found that the Russians and the Japanese are not adverse to robotics in whatever permutation. The so-called fear comes from America of all places.
Nice punchline.

Bostrom argues that a super intelligence (unless programmed extremely carefully) will inevitably seek dominance so that it can accomplish its goals more efficiently. But this idea seems to come from his very specific notion of superintelligence as an (ironically) rather dumb but unimaginably powerful optimiser. This model, in turn, comes from the idea that the quickest way to build a superintelligent AI would be for a less intelligent one to improve on its own design. Superintelligence would come at the end of long chain of "recursive self-improvement" cycles and we humans would only need to design the first, not particularly intelligent agent to get the whole thing started.

To me this seems way too speculative and I'd like to see some real examples of this sort of unconstrained self-improvement before I take it seriously as a technology. Until then it's just ingenious science fiction. But it seems to have convinced a lot of people.
YOUR COMMENT Login or Register to post a comment.

Next entry: Liberalism’s Great Challenge: How Can We Critique Ideas while Protecting People?

Previous entry: Our emerging culture of shame