EDIT: I linked to XKCD here... cuz never miss an opportunity for xkcd! Also, I just finished reading the article, so I'll add my comments on the content.
The parent post (for this comment) contains a rant about physicists but has no specific rebuttal of the article.
David Deutsch's arguments seem very sensible to me, in a completely non-mystical sort of way. I feel he asks questions pointing to the heart of AI. Regarding moving goalposts and stuff, you might like to read Freeman Dyson's (excellent and insightful) article on 'Birds and frogs' where he comments on such things. Deutsch is approaching the problem like a bird, typical AI researchers are approaching the problem in a frog-like manner.
And oh, to state possible biases on record: I'm a physicist :-)
I actually catch myself sometimes saying or thinking the tooltip text. :)
But what Deutsch writes here is somewhat opposite in approach to the problem of how to be annoying--he's trying make a non-physics research area seem fundamentally more complicated than it likely is. I'm willing to say it borders on mysticism, and the only reason I can think of for such behaviour in an otherwise fine scientist is a strong emotional attachment to some romantic idea of what it is to be a human being. There is zero (AFAIK) theoretical or empirical evidence for his grand claims about creativity, and he's playing the old game in AI: goalpost moving. Whenever AI does something that was previously thought to be only achievable by humans, the critics immediately jump and say: "Ah, but that's not really intelligent, is it?"
> The parent post (for this comment) contains a rant about physicists but has no specific rebuttal of the article.
It's not a rant about physicists (I am one, for the record), it's just that this is what I noticed--whenever there's an otherwise good scientist giving misguided proclamations on how some research field other than his should proceed, it's always the physicist. Maybe I'm biased in that I take extra notice of what physicists say, but anyway it's anegdotal and not the main point. What matters is: whoever you are, if you're going to seriously opine on a subject, beyond general methodological remarks (which are extremely important), you better have in depth knowledge of the subject matter. Example claim: "behaviourism is abandoned by mainstream psychology". I'll avoid discussing what he means by "mainstream" here (there could be all sorts of ugly dragons lurking behind this statement), and just say that this is false. The term behaviourism is not used anymore (it's not useful anymore to use it), but a huge deal of behavioural theory and results are still alive and well in psychology of cognition.
His central point, that "creativity" is the key to AGI, is meaningless the way he talks about it. Creative thinking is not a mysterious process, there is a lot of empirical work dealing with it. He further postulates that we need philosophical (of all things) breakthroughs. I can't help but think this is some sort of Chinese room style argument all over again. Having stated, multiple times, that philosophical/epistemological breakthroughs are needed, he speculates that maybe better understanding of genetic diffs between humans and other higher primates hold the key. It's all over the place. It's hard to give specific rebuttals to extremely vague ideas.
Thanks for the pointer to Dyson's article, found it and stashed it for later reading.
> And oh, to state possible biases on record: I'm a physicist :-)
Have you read Chesterton's The Man Who Was Thursday? This discussion is starting to feel like it, just do a %s/anarchists/physicists/g :)
EDIT: I linked to XKCD here... cuz never miss an opportunity for xkcd! Also, I just finished reading the article, so I'll add my comments on the content.
The parent post (for this comment) contains a rant about physicists but has no specific rebuttal of the article. David Deutsch's arguments seem very sensible to me, in a completely non-mystical sort of way. I feel he asks questions pointing to the heart of AI. Regarding moving goalposts and stuff, you might like to read Freeman Dyson's (excellent and insightful) article on 'Birds and frogs' where he comments on such things. Deutsch is approaching the problem like a bird, typical AI researchers are approaching the problem in a frog-like manner.
And oh, to state possible biases on record: I'm a physicist :-)