Thursday, November 20, 2008

Trope-ing the light fantastic (AI)

Artificial intelligence is another old favorite of SF authors. Using ourselves as proof by example, intelligence is possible. So why not artificial intelligence? AI feasibility seems like a (pardon the expression) no-brainer -- except for the fact the research community has been on the verge of AI since computers were invented.

Maybe it would help if we could agree what intelligence, awareness, or consciousness mean. In the absence of agreement, we fall back on the Turing Test. We'll have AI when we can't tell an AI from a human on the other end of a comm line. Like beauty and obscenity, intelligence is in the eye of the beholder.

Perhaps our anthropomorphism is the problem. As the hero in my book Fools' Experiments opines of the Turing Test:

What kind of criterion was that? Human languages were morasses of homonyms and synonyms, dialects and slang, moods and cases and irregular verbs. Human language shifted over time, often for no better reason than that people could not be bothered to enunciate. "I could care less" and "I couldn’t care less" somehow meant the same thing. If researchers weren't so anthropomorphic in their thinking, maybe the world would have AI. Any reasoning creature would take one look at natural language and question human intelligence.

I don't think AI is a trope -- just that we need to approach the problem a different way.

2 comments:

AReichl said...

Hello,

normally answering to a post is easier than beginning one, but let's try.

1)

In my opinion the Turing Test is NOT enaugh.

Even if the "discussion" with the "entity" on the other side is satisfactory (hey - that's a nice guy; and sooo friendly), you will never know, if it's a [grown-up] intelligent mind (human,alien,gene-enhanced chimpanzee) TALKING to you or a [very good] program MAKING YOU THINK it would talk to you.

If there ever will be such programs, they will be usefull, but they probably never will get civil rights. Of course we will develop a kind of relationship - kids will weep if one is switched off - but that is another story.

2)

If the brain is working more like an analog computer or even a quantum computer or a mixture of both, you cannot simulate it with a digital one (maybe with a quantum computer).

The point for understanding the brain is the following question:

Can you take a "snapshot"?

(What you do with it is not important here. Maybe you do it just in case something happens to you - grow a duplicate body, put back the snapshot and continue life from there on).

If you make a photo of a passing car, you don't know its speed or direction (motion blurr aside).
Same with the brain. If thoughts are the CHANGE in chemistry or electrical potentials then it would have to be some kind of "dynamic" snapshot.

A digital computer can be stopped. Now. Make a snapshot (in Windows its "Hibernate"), switch off, switch on, put back snapshot and continue. Just Bits and Bytes, no mind or soul.

Edward M. Lerner said...

Agreed: The Turing test isn't *the* definitive test -- it's too anthropocentric. There's no reason to believe humans are the only possible model for intelligence.

(That said, I consider the Turing test to be *a* valid standard. Failing it can't disprove intelligence, but passing it shows human-style intelligence.)

Can one take a snapshot of an intelligent mind? You propose not. I'd like to hear more about how you reached that conclusion.