"Intelligent Machinery" by A.M.Turing | Hacker News Reader

"Intelligent Machinery" by A.M.Turing

alanturing.net

40 pointsColinWright13 comments

I think the realization that humans are merely computational systems is one of the more important parts of my life.

I don't think 'realization' is really the right word unless you have proof.

I don't have evidence to suggest anything else.

My previous theory on human intelligence involved a supernatural belief. Any other alternative that I can conceive of, besides the computation theory, similarly rely on supernatural beliefs.

I reject the assumed suggestion that I must have evidence to reject the supernatural. Or, in simpler terms, dualism requires evidence while monism does not (although it of course has it, in great quantity).

All I'm saying is that in biological sciences, exactly how we "work" (and in particular, how our brains work) is still very much an open question, just as whether a machine can have intelligence is an open question in computer science. Unfortunately not having evidence for other theories does not even come close to proof of the theory in question.

As long as all of reality runs on mathematics then AI is possible. I think many commonly assume that reality is inherently mathematical, but don't realize their assumption.

The question isn't "how do they work". The question is "are they more computationally powerful than Turing machines". Until demonstrated otherwise, the only sound position is "No."

I've often been stumped why the notion of "strong AI" cannot be successfully refuted. On the one hand any analog system can be modeled to arbitrary precision by a digital one; and on the other hand it is also a fallacy to mistake a model for the thing itself (at least from the perspective of Aristotelian logic). The notion consciousness just arises at some level of complexity is a deus ex machina expectation, and not scientific, but this is not the argument the best proponents of strong AI put forward. Instead, I think the clincher argument itself points out the limitations of science. Science is based on observation. If intelligence is an observed phenomena (and I'm still not satisfied we have an adequate definition; it seems more of an I know it when I see it definition) then clearly, machines can be intelligent (but conscious !!??).

Just because something is an observed phenomena doesn't mean it is the result of mathematical laws and therefore replicable by a computer. While that may be the most acceptable position in today's intellectual climate, it isn't a logical necessity.

Unless the universe contains mechanisms to allow more powerful computation models than we are aware of, then I cannot see how it could possibly not be possible.

Intelligence is observed in nature in little pieces of meat that arose through natural selection. The notion that it is doing anything that possibly couldn't fall under the umbrella of computation is one that does not logically make sense. The only logical position to assume is that until brains can be demonstrated to be solving problems that Turing machines cannot, they must be assumed to be computationally equivalent.

No, in the realm of logical necessity, there is nothing necessitating observed phenomena be the product of computations. I could state this in terms of Kolmogorov complexity: there could be specific events with higher KC than the origin of the universe. I don't see anything inherently incoherent about that statement.

It doesn't fit within modern deterministic/stochastic metaphysics, but that's besides the point.

Anyone have a more readable version of this?

Also a classic -- http://www.jstor.org/stable/pdfplus/2251299.pdf?acceptTC=tru...

And that same classic, not behind a paywall -- http://www.loebner.net/Prizef/TuringArticle.html

Reply on news.ycombinator.com