James Barat seemed like a nice enough guy, but even though i wouldn’t like to see him upset, the review seems a bit too uncritical. His arguments, if not false, had some pretty obvious weaknesses — made more obvious by said arguments endless repetition — that you might have pointed out. …. (Of course the obvious response to this is “write your own damn review”, and fair enough)
I was very happy that Barrat wrote the long-awaited summary of the intelligence explosion concept. Kurzweillians have The Singularity is Near; there is no equivalent for MIRI.
There is a need for well-known “standard” intros to point people to when they first hear of the concept and want to learn more. Ideally, there would be items in various lengths, e.g, 200 words, 1500, 20000, and full book length.
Barrat’s book is is suitable for general readers. But for the sort of super-smart person I would want to recruit as a MIRI support, it lacks the rigor or clarity of some in-house MIRI publications.
For them, I’d recommend Staring into the Intelligence Explosion as in introduction, though I think that more is still needed.
“Instead Barrat faults Kurzweil and others like him for trumpeting AI’s promise while minimizing its peril…”
This exactly mirrors my reaction toward Kurzweil et al. It’s not so much that he essentially rubs his temples and intones wildly-oversold predictions like papal decrees…or that he trips over his shoelaces when broaching topics like human nutrition or neuroscience (of which he is little more than a self-professed “expert”). It’s the rapturous way he describes the “Singularity,” the way he glosses over the possibility that the birth of super-human intelligence might coincide with our extinction, that no malevolence is required for a strong AI to merely see us as dispensable in the context of goals that differ from our own (goals that we may not be able to comprehend).
The military research aspect mentioned in the book is a cogent reminder of just what kind of AI might “get there first.”
The writing on the wall makes it difficult to understand why people generally don’t see the need for research like that undertaken by MIRI.
6 Responses to “Will Advanced AI Be Our Final Invention?”
December 15
Michael WitbrockJames Barat seemed like a nice enough guy, but even though i wouldn’t like to see him upset, the review seems a bit too uncritical. His arguments, if not false, had some pretty obvious weaknesses — made more obvious by said arguments endless repetition — that you might have pointed out. …. (Of course the obvious response to this is “write your own damn review”, and fair enough)
December 15
Joshua FoxI was very happy that Barrat wrote the long-awaited summary of the intelligence explosion concept. Kurzweillians have The Singularity is Near; there is no equivalent for MIRI.
There is a need for well-known “standard” intros to point people to when they first hear of the concept and want to learn more. Ideally, there would be items in various lengths, e.g, 200 words, 1500, 20000, and full book length.
Barrat’s book is is suitable for general readers. But for the sort of super-smart person I would want to recruit as a MIRI support, it lacks the rigor or clarity of some in-house MIRI publications.
For them, I’d recommend Staring into the Intelligence Explosion as in introduction, though I think that more is still needed.
December 15
Marius van VoordenReview link doesn’t work for me. Try this one. http://www.amazon.com/gp/cdp/member-reviews/A1SD50PZZJ8YU1/
December 15
Louie HelmI did also write an Amazon review. The one I like to above is here though http://singularityhub.com/2013/12/14/will-advanced-ai-be-our-final-invention/
December 16
Marius van VoordenOh right, seems I didn’t read properly and thought the link that led to Amazon should have led to your review.
December 16
Taurus“Instead Barrat faults Kurzweil and others like him for trumpeting AI’s promise while minimizing its peril…”
This exactly mirrors my reaction toward Kurzweil et al. It’s not so much that he essentially rubs his temples and intones wildly-oversold predictions like papal decrees…or that he trips over his shoelaces when broaching topics like human nutrition or neuroscience (of which he is little more than a self-professed “expert”). It’s the rapturous way he describes the “Singularity,” the way he glosses over the possibility that the birth of super-human intelligence might coincide with our extinction, that no malevolence is required for a strong AI to merely see us as dispensable in the context of goals that differ from our own (goals that we may not be able to comprehend).
The military research aspect mentioned in the book is a cogent reminder of just what kind of AI might “get there first.”
The writing on the wall makes it difficult to understand why people generally don’t see the need for research like that undertaken by MIRI.