You could see Miller's _Singularity Rising_ as an attempt to swim against the book current of Ray Kurzweil and present some of the *other* visions of the Singularity: specifically, the Intelligence Explosion school as exemplified by Eliezer Yudkowsky and Robin Hanson. It then mixes in a bunch of material on intelligence & genetics, so we might identify an additional subschool: that of Steve Hsu on embryo selection for increasing human intelligence.

Miller succeeds in giving a wide overview of quite a few topics, from Hanson's 'crack of a future dawn' em scenario to the Great Filter to comparative advantage & the advantages of trade as it applies (and doesn't apply) to AIs to the intelligence orthogonality thesis (that intelligence does not imply benevolence) to the logic of arms race and its particularly unpleasant applicability to AI development. And then he tosses in the mentioned intelligence & genetics material, which I was a little surprised to learn from - I had read many of his citations (and actually host a few of the online copies of the papers on my personal site, gwern.net), but he still threw in some ones that were new to me.

On a purely factual basis, I have relatively little to fault Miller for. He makes a risible claim about 1700s French life expectancies not hitting the 50s (true only if you include infant mortality, otherwise hitting 50s was perfectly routine - even in the worst tabulations, generally if you made it to 20 on average you would reach the 50s; see 0, 1, 2, 3, 4) but he is far from the first to make that mistake; he brings up dual n-back more than once, but he avoids making too many or overreaching claims on behalf of dual n-back such as the increasingly questionable effect on intelligence (see my meta-analysis); he seems to criticize people for not taking seriously the method of castration for life extension but doesn't mention the issues with the data and the likelihood that the method would not work post-puberty (ie. for everyone who is able to morally consent to such a procedure). Otherwise...

Otherwise Miller's sins are simply that the writing is merely OK and while he does a reasonable job of, as Hanson puts it in his own review of _Singularity Rising_, "explaining common positions and intuitions behind common arguments", he barely defends them or clearly justifies them. While I and many others involved in the area dislike Ray Kurzweil's theories and arguments and books as being superficial, right for the wrong reason, overly optimistic etc, they do at least do their job of convincing people (and then hopefully they can adopt more nuanced or different views); but though I agree with a large fraction of it, it's hard to believe that anyone could read Miller's book and come out genuinely convinced of pretty much anything in it (as opposed to reactions like "that's interesting" or "maybe"). For example, he does a nice question-answer sequence against the kneejerk bad-philosophy reactions to cryonics, but one could easily bite all the bullets and simply question the incredibly sketchy case he makes (yes, it's great that wood frogs do cryonics all the time, but we're not frogs). He asks that anyone who signs up for cryonics email him about what convinced them - I immediately thought, "50% odds that no one has done so yet". (After writing this review, I asked Miller about this and he said no one had yet.)

And aside from as comprehensive a layman discussion of the issues involved in AI economics and technological unemployment as I've ever seen, I can't really name any original contribution this book makes.

I can't say I'm really glad I read it, but then I can't say I really regret reading it (I got a number of IQ-related citations, a discussion of neo-Luddism, and info on the more esoteric possibilities of embryo selection). This is because I already know almost everything in the book and have read many of the citations already, so I am not the target audience; it's good if you want an overview of non-Kurzweilian Singularity ideas and you don't want to read through scores of webpages and papers, and more or less unique in conveying them all in a compact single place - so in acknowledgment of this, I bump my rating up to 4 stars (though for me it was more like 3).

Excerpts:

- intro-ch3
- ch4-5
- ch9
- ch10-12
- ch13
- ch14-15
- ch16