4 ms·

Fair play for machines: Chomsky's misreading of Turing, and why it matters

31 points by djestrada 13y ago

juliangamble 13y ago
The conclusion is a pleasant antidote to the title (which to its credit got me to read it). >It's time we all play fair. Turing and Chomsky are both giants, but its a mistake to suggest they're operating in the same field. This mistake is also made by people who read the 'conflict' between Norvig and Chomsky as saying anything about the future of AI: http://norvig.com/chomsky.html http://norvig.com/chomsky.html http://www.tor.com/blogs/2011/06/norvig-vs-chomsky-and-the-fight-for-the-future-of-ai http://www.tor.com/blogs/2011/06/norvig-vs-chomsky-and-the-f... http://www.theatlantic.com/technology/archive/2012/11/noam-chomsky-on-where-artificial-intelligence-went-wrong/261637/ http://www.theatlantic.com/technology/archive/2012/11/noam-c... (Norvig is also a giant - but he is at least honest about the limitations and scope of his work). Turing was essentially speaking to his context at the time: > Turing says the question "can machines think" isn't helpful in this process because it invokes conceptual and prejudicial biases about "thinking" and "machines" that themselves can't be clearly resolved, largely because they rely on confused metaphysical (or superstitious) beliefs. Turing's claim of "meaninglessness" reflects the positivist attitude of the scientists and logicians that formed Turing's intellectual community. Chomsky is coming from a philosophical and linguistic background and saying that AI hasn't gone as far as he hoped it would: > Chomsky has a generally very pessimistic view of AI and he's trying to enlist Turing on his side but it's simply an abuse of the view. Chomsky also claims in this interview that the last 40 years of AI research "hasn't given any insight to speak of into the nature of thought and organization of action". For some really awesome stuff in terms of the future of AI - the person to look at is Jeff Hawkins: http://www.ted.com/talks/jeff_hawkins_on_how_brain_science_will_change_computing.html http://www.ted.com/talks/jeff_hawkins_on_how_brain_science_w... http://www.infoq.com/presentations/Brain-Computing http://www.infoq.com/presentations/Brain-Computing I'm massively oversimplifying - but it looks like his group has built some products by modelling a Cortical Learning Algorithm. In a Computer Science sense - this looks like a bitset corresponding to memories of facts - that you can then 'diff' - to make higher level judgments. The example he gave was groups of idea relations on wikipedia. Taking the ideas for 'Apple' and subtracting the ideas for 'Fruit' - left you with the ideas for 'Computer'.
- p1esk 13y ago
  Norvig claims you can make progress in AI using probability theory and statistical analysis (IBM Watson style), and Chomsky is clearly skeptical about that approach. Seems like a conflict to me. I do like Hawkins' ideas, but so far neither he, nor Dileep George has made any significant breakthroughs, and it's been 10 years since "On Intelligence" was published.
yetanotherphd 13y ago
I'm not an expert in AI or NLP, but my impression is that Chomsky is opposed to any progress in formal semantics, probably because it goes beyond the formal syntax that he worked on. Luckily everyone in the field is free to ignore him and work on it anyway. There was a beautiful attempt by Roger Schank and his students to come up with a better mathematical description of human reasoning. This fizzled out without making much progress, but who knows what the future holds.
eruditely 13y ago
This article was incredible. A clear and simple handling of a difficult subject. Brilliant.
- djestrada 13y ago
  Thanks, I appreciate it!
calibraxis 13y ago
I enjoyed this article. But to clarify Chomsky's claim, he once used the analogy of submarines. In some languages (Russian, I vaguely recall?), submarines swim. In English, submarines don't swim. It's a matter of language. Any of us would be willing to accept the analogy, unless we're being really pedantic; it's not a big deal. (I personally used the analogy last week, when I said, "I believe that was 2011, but maybe your computer thinks otherwise?" and the (non-IT) person immediately understood without comment.)
- djestrada 13y ago
  Jay Gordon defends Chomsky's view more thoroughly against my criticism here: https://plus.google.com/110181959253415840755/posts/7y2s5e9Sggk https://plus.google.com/110181959253415840755/posts/7y2s5e9S... I also leave some responses on that post.
  - calibraxis 13y ago
    Thank you! A little off-topic, I like the idea of seeing AI research as an industry, and looking at institutional problems which make its output worse. Anthropologist David Graeber discusses this in (http://thebaffler.com/past/of_flying_cars http://thebaffler.com/past/of_flying_cars) and (https://vimeo.com/41863599 https://vimeo.com/41863599). (The article and video focus on different things; the article's editors removed some arguments about bureaucratic technologies — on the internet, we fill out forms all day — and the video delves more into that.)

31 points by djestrada 13y ago

The conclusion is a pleasant antidote to the title (which to its credit got me to read it). >It's time we all play fair. Turing and Chomsky are both giants, but its a mistake to suggest they're operating in the same field. This mistake is also made by people who read the 'conflict' between Norvig and Chomsky as saying anything about the future of AI: http://norvig.com/chomsky.html http://norvig.com/chomsky.html http://www.tor.com/blogs/2011/06/norvig-vs-chomsky-and-the-fight-for-the-future-of-ai http://www.tor.com/blogs/2011/06/norvig-vs-chomsky-and-the-f... http://www.theatlantic.com/technology/archive/2012/11/noam-chomsky-on-where-artificial-intelligence-went-wrong/261637/ http://www.theatlantic.com/technology/archive/2012/11/noam-c... (Norvig is also a giant - but he is at least honest about the limitations and scope of his work). Turing was essentially speaking to his context at the time: > Turing says the question "can machines think" isn't helpful in this process because it invokes conceptual and prejudicial biases about "thinking" and "machines" that themselves can't be clearly resolved, largely because they rely on confused metaphysical (or superstitious) beliefs. Turing's claim of "meaninglessness" reflects the positivist attitude of the scientists and logicians that formed Turing's intellectual community. Chomsky is coming from a philosophical and linguistic background and saying that AI hasn't gone as far as he hoped it would: > Chomsky has a generally very pessimistic view of AI and he's trying to enlist Turing on his side but it's simply an abuse of the view. Chomsky also claims in this interview that the last 40 years of AI research "hasn't given any insight to speak of into the nature of thought and organization of action". For some really awesome stuff in terms of the future of AI - the person to look at is Jeff Hawkins: http://www.ted.com/talks/jeff_hawkins_on_how_brain_science_will_change_computing.html http://www.ted.com/talks/jeff_hawkins_on_how_brain_science_w... http://www.infoq.com/presentations/Brain-Computing http://www.infoq.com/presentations/Brain-Computing I'm massively oversimplifying - but it looks like his group has built some products by modelling a Cortical Learning Algorithm. In a Computer Science sense - this looks like a bitset corresponding to memories of facts - that you can then 'diff' - to make higher level judgments. The example he gave was groups of idea relations on wikipedia. Taking the ideas for 'Apple' and subtracting the ideas for 'Fruit' - left you with the ideas for 'Computer'.

p1esk 13y ago
Norvig claims you can make progress in AI using probability theory and statistical analysis (IBM Watson style), and Chomsky is clearly skeptical about that approach. Seems like a conflict to me. I do like Hawkins' ideas, but so far neither he, nor Dileep George has made any significant breakthroughs, and it's been 10 years since "On Intelligence" was published.

yetanotherphd 13y ago

I'm not an expert in AI or NLP, but my impression is that Chomsky is opposed to any progress in formal semantics, probably because it goes beyond the formal syntax that he worked on. Luckily everyone in the field is free to ignore him and work on it anyway. There was a beautiful attempt by Roger Schank and his students to come up with a better mathematical description of human reasoning. This fizzled out without making much progress, but who knows what the future holds.

eruditely 13y ago

This article was incredible. A clear and simple handling of a difficult subject. Brilliant.

djestrada 13y ago
Thanks, I appreciate it!

calibraxis 13y ago

I enjoyed this article. But to clarify Chomsky's claim, he once used the analogy of submarines. In some languages (Russian, I vaguely recall?), submarines swim. In English, submarines don't swim. It's a matter of language. Any of us would be willing to accept the analogy, unless we're being really pedantic; it's not a big deal. (I personally used the analogy last week, when I said, "I believe that was 2011, but maybe your computer thinks otherwise?" and the (non-IT) person immediately understood without comment.)

djestrada 13y ago
Jay Gordon defends Chomsky's view more thoroughly against my criticism here: https://plus.google.com/110181959253415840755/posts/7y2s5e9Sggk https://plus.google.com/110181959253415840755/posts/7y2s5e9S... I also leave some responses on that post.
- calibraxis 13y ago
  Thank you! A little off-topic, I like the idea of seeing AI research as an industry, and looking at institutional problems which make its output worse. Anthropologist David Graeber discusses this in (http://thebaffler.com/past/of_flying_cars http://thebaffler.com/past/of_flying_cars) and (https://vimeo.com/41863599 https://vimeo.com/41863599). (The article and video focus on different things; the article's editors removed some arguments about bureaucratic technologies — on the internet, we fill out forms all day — and the video delves more into that.)