4 ms·

Yudkowsky and Soares' Book Is Lacking (2025)

11 points by measurablefunc 2mo ago

functionmouse 2mo ago
> they don’t present any evidence, or discuss how computers might get to intelligence. Rationalists don't need pesky things like data or evidence to assert their desired conclusions as fact. Don't you know, you can assert your vibes as science, now.
lern_too_spel 2mo ago
Is this an AI bot writing about why AI cannot exist like some new kind of AI performance art? What are we looking at here?
- tom_ 2mo ago
  It read to me like a book review written by a person. What do you see as the bot tells? I didn't spot the usual obvious things, and by the time I got to the end I wasn't feeling vaguely suspicious. (But, if this is actually AI-written, and possibly quite transparently so, then I've clearly missed something, and I'd quite like to know what!)
  - lern_too_spel 2mo ago
    It was so poorly written that I assumed there was some other reason for it being posted than the content of the article.
benlivengood 2mo ago
If we had a theory of intelligence that allowed us to say "this is why and how machines are intelligent and how they become more intelligent" we'd likely be very close to solving the alignment problem as well, obviating the need for the book. The book fits into the current unknown-unknowns world of rapidly increasing machine-learning capabilities across most domains where there is little evidence to suggest that "intelligence" is limited to the human limit (which is already quite high relative to the 99.9th percentile; we get a few Einstein, Von Neumann, Turing class people in a century) or the human body.
fwlr 2mo ago
So, they say, computers are not smart now, we don’t know what intelligence is, we don’t know how to make computers smart, but they certainly will be intelligent soon. There is currently a trillion-dollar corporation whose valuation largely rests on their progress in solving this. It is beyond asinine to require it from a book, as though it were table stakes for discussion.

> they don’t present any evidence, or discuss how computers might get to intelligence. Rationalists don't need pesky things like data or evidence to assert their desired conclusions as fact. Don't you know, you can assert your vibes as science, now.

lern_too_spel 2mo ago

Is this an AI bot writing about why AI cannot exist like some new kind of AI performance art? What are we looking at here?

tom_ 2mo ago
It read to me like a book review written by a person. What do you see as the bot tells? I didn't spot the usual obvious things, and by the time I got to the end I wasn't feeling vaguely suspicious. (But, if this is actually AI-written, and possibly quite transparently so, then I've clearly missed something, and I'd quite like to know what!)
- lern_too_spel 2mo ago
  It was so poorly written that I assumed there was some other reason for it being posted than the content of the article.

benlivengood 2mo ago

If we had a theory of intelligence that allowed us to say "this is why and how machines are intelligent and how they become more intelligent" we'd likely be very close to solving the alignment problem as well, obviating the need for the book. The book fits into the current unknown-unknowns world of rapidly increasing machine-learning capabilities across most domains where there is little evidence to suggest that "intelligence" is limited to the human limit (which is already quite high relative to the 99.9th percentile; we get a few Einstein, Von Neumann, Turing class people in a century) or the human body.

fwlr 2mo ago

So, they say, computers are not smart now, we don’t know what intelligence is, we don’t know how to make computers smart, but they certainly will be intelligent soon. There is currently a trillion-dollar corporation whose valuation largely rests on their progress in solving this. It is beyond asinine to require it from a book, as though it were table stakes for discussion.