Superintelligence: Paths, Dangers, Strategies

This book has had me intrigued for a while now. It’s all about the potential advent of strong AI, and hypothesizes some future scenarios regarding what sort of form such “superintelligence” might take, and what should happen if or when such sudden ascendance comes to pass. It’s “profoundly ambitious and original…nothing less than a reconceptualization of the essential task of our time.”

Bostrom takes us from past developments and present capabilities of AI, to various potential paths to superintelligence, to the forms this superintelligence might take and the powers and advantages such forms might confer, to the potential scenarios and problems that we (humans!) might face as a result, and finally to some strategies with which we might approach this—ideas for prolonging our species, for our end game, if you will.

This looks to be a provocative book on an inherently fascinating topic, timely and relevant in a lot of ways, and fairly accessible—I’d say there’s a pretty high chance this is one I’ll eventually get around to reading. (Even though apparently the style can be “dense and dry”…)

A lot of what this book touches on is, of course, highly speculative. There seem to be a couple camps when it comes to AI research: most people find the question of strong AI and the possibility of a “singularity” interesting, if a farfetched distant future; some, though, see it as a tremendously critical and urgent existential threat—indeed as the major, defining issue of our time.

From my perspective, sitting here, small-brained organic flesh-blob as I am, thinking my limited thoughts, it does seems a bit out there. But a lot of really smart and thoughtful people agree that we should take it very seriously, so I am at the very least willing to entertain these scenarios. Besides, regardless of how likely they actually are, the fascinating philosophical futurist questions such a book stimulates can only serve to make us more mindful about how we steward our collective future.