John Searle

[ciphergoth]

The question I asked was

"The Church-Turing thesis states that any machine we can really imagine building, certainly any machine that can be built using the physical law that we know, can be simulated on a computer. That includes the human brain, which we agree is a machine. So do you agree with Penrose that there's physical law we don't know that will extend the powers of the brain beyond those of a Turing machine?"

The question I should have asked was

"When you represent what you call Strong AI as being based on the belief that the brain is like a digital computer, that's a deliberate misrepresentation designed to make it seem less plausible. Strong AI is, as you know perfectly well, based on the belief (which you share) that the brain is some sort of machine, and as such is amenable to simulation on a computer. You fuck."

As for the way he misrepresents Dennett... well, anyway, I'm kicking myself because I'll never get the chance again...

Dreamt about a Goth weekend being run by this year's BiCon committee, in a town a bit like Whitby but different; people were bemoaning the absence of the Elsinore. Instead of sleeping in beds, we slept in mattress-shaped tanks of water; they were quite comfy once the heat of your body warmed the water. I went to what I thought was a plenary, but it turned out to be a crisis meeting of the commitee; adjectivemarcus said I should stay because of being involved in last year. They were trying to call emergency services because of some sort of drug-related medical emergency, but their mobiles weren't getting reception and the landline was tied up because (babysimon explained to me) some interfering busybody had insisted that the best way to get them would be to dial out and raise them online...

Comments

[babysimon]

> The question I should have asked was...

Am I being dumb here? What's the difference between "like a digital computer" and "amenable to being simulated on a digital computer"?

> Instead of sleeping in beds, we slept in mattress-shaped tanks of water

That's supposed to be a surprise!

[ciphergoth.livejournal.com]

Ask Searle - he believes the latter but not the former.

But also "like a digital computer" could suggest an image of things like "works in binary" or "has a clock cycle" or "is designed to be very deterministic and predicable" or all sorts of things that are true of real computers in the real world but not of brains. To put it another way, by that interpretation everything in the world is like a digital computer, which makes it mean little.

[babysimon]

Ah. In that case, I don't think that was the question you should have asked either :)

Maybe you should have asked him why he isn't a proponent of strong AI given that (you say) he believes all the right stuff...

[lovingboth]

Does Church-Turing say that? When Turing talked of 'mechanical', didn't he mean something more like 'without thought or insight'?

For a start, surely there are machines we can build that we cannot accurately simulate. A certain radioactivity-based random number generator springs to mind. If you could accurately simulate it, it wouldn't be random...

[lovingboth]

http://www.alanturing.net/turing_archive/archive/m/m01/M01-005.html would appear to be an example of what he's talking about: something a human could do with enough paper, pencils and instructions to blindly follow, a Turing Machine can do too.

(Apart from pick its nose with one of the pencils etc!)

[ciphergoth.livejournal.com]

In formal settings, you sometimes see it defined as "a Turing machine coupled with a source of true random bits".

C-T extends to analogue systems too, though I don't know the proof, and as far as the system defined by the sort of physical law we currently think we have.

[ciphergoth.livejournal.com]

He did answer that question, and his answer is weird.

He says "a simulation of digestion isn't real digestion, and a simulation of thought isn't real thought. A machine may be able to provide a perfect simulacrum of thinking in every respect, but it isn't really thinking; it's just simulated thinking."

It makes me think of the idea that some subset of humanity don't have souls...

[babysimon]

> a simulation of digestion isn't real digestion, and a simulation of thought isn't real thought

...and of course, digestion and thought are similar enough concepts that the analogy works...

[lovingboth]

ie a machine we can't simulate connected to a TM.

I'm not convinced by the analogue one either: we can't simulate - again, in the rather important form of being able to predict the answer - one of those magnet on a wire moving around some fixed magnets before settling in a stable position 'decision maker' executive toys.

[ex-meta.livejournal.com]

A simulation of adding numbers doesn't actually add numbers, naturally. Only I can do that with my holy brain.

[ciphergoth.livejournal.com]

This is effectively the randomness thing in another guise. We can't predict a die roll, but we can simulate one. There's an important sense in which a computer can perform any computational task any analogue system which includes noise can perform.

OMAS

[some-fox.livejournal.com]

In asking Searle a question you may have stumbled up against OMAS (older male academic syndrome), a viralent strain of which seems to infect all men (and even some women) who have been in the academic world for more than a couple of years. Warning signs are:
- expressing simple concepts in unnecessarily complicated ways in order to alienate and confuse mere mortals
- answering questions with irrelevant statements about something they'd rather talk about
- ignoring points made by young/female/alternative-looking people and then applauding the Exact Same Point when it is made by another OMA
- a complete inability to confront the possibility that they might be (shock, horror)... wrong.
Sounds like he has an incurable case. The only way forward is to hurl Searle into the OMAS pit and leave him there with all the other pompous, arrogant gits.

Btw, where do you stand on the connectionist view of the brain? From my limited understanding of such things it seems to make a lot of sense and there seems no reason why this way of brain functioning couldn't potentially be modelled on a machine.

Re: OMAS

[some-fox.livejournal.com]

Oops - I meant 'most' men, not 'all' men of course. Some, happily, seem to be immune.

Oh, and another symptom of OMAS is asking a question that's not really a question but a statement that enables them to talk about their own ideas. Grr.

Re: OMAS

[pavlos.livejournal.com]

Although I don't doubt the existence of this syndrome, I think Searle is making a straightforward AFPA (Argument From Personal Incredulity): "It seems really implausible to me, so it can't be true."

Pavlos

Fear my supercomputer!

[pavlos.livejournal.com]

A simulation of a nuclear strike is not a nuclear strike. A simulation of a nuclear test is a nuclear test. No philosophical paradox here, the only difference is that the real nuclear strike, as for real digestion, affects specific atoms that you have a vested interest in.

Really, I do find this Searle fellow rather dumb. Or rather, I think he's asking an important question such as "At what point does a complex entity acquire perception/feelings/self-perception/whatever?" and he's asking it in an extraordinarily dumb way.

Pavlos

Re: AFPI

[some-fox.livejournal.com]

Oh it's a related disease to be sure :-)

Re: OMAS

[thekumquat.livejournal.com]

- as described by Jonathan Slack in his book 'From Egg to Ego' - how it is near-impossible to get on in academia without being an arrogant tosser, and including himself in that. I believe he mentioned having to fit in, ie be an OMA as part of it.

[the title is a pun after his classic embryology text From Egg to Embryo]
Must update Amazon wishlist...

Re: OMAS

[ciphergoth.livejournal.com]

Hmm, I know the syndrome but I don't think it accurately characterises Searle's problem...

Re: OMAS

[some-fox.livejournal.com]

Hm. Seems I'm not being too successful at responding to your LJs. Ah well.

[lovingboth]

Up to a point, Lord Copper.

True dice are random. How do computers produce random (as opposed to psuedo-random) numbers? They can't - humans have to go off and make some physical hardware (like Walker's radioactive decay box or ERNIE's electronics) that does it for them.

It is true that it's possible to simulate a roulette wheel in real time with sufficient accuracy to make money (I'm more amazed that the people wot did it managed it with a 6502 programmed in hex...) albeit at some risk. But what that tells us is that roulette wheels aren't random!

And "any machine" is still way over what Turing in particular actually claimed.

'Thesis M' is such an extension: Whatever can be calculated by a machine (working on finite data in accordance with a finite program of instructions) is Turing-machine-computable.

And that's unproven, even if constrained to machines that could actually exist in the real world.

What I sense you'd like to have proven is Thesis M with the bit in parenthesis taken out... and I don't think that is ever going to happen.