Re: >H Re: Let's work on the FAQ!

Michael Nielsen (
Sun, 12 Jul 1998 12:59:42 -0600 (MDT)

On Sun, 12 Jul 1998 wrote:

> Michael Nielsen writes:
> > Transhuman Mailing List
> > > What is nanotechnology?
> > What is artificial intelligence?
> I think we could stand some internal debate on this subject, right
> now. I've always had a side interest in artificial intelligence and
> artificial cognition, but have never had the time to do any serious
> reading on the topic.
> My clever layman's impression of the subject, however, is that no two
> afficianados (and very few actual researchers) have progressed to the
> point of common definitions for the fundamental topics.

That is also my impression.

> I may also be an iconoclast, but I've also never ever seen the point
> of the Turing test (as I understand it) and Greg Egan's _Permutation
> City_ did nothing for me except to convince me that something is
> desperately missing from that idea of digital consciousness.

Passing the Turing Test seems to me to be a sufficient (though not necessary) condition for intelligence. If a machine can jump through all the hoops which I am able to conure up, then I would consider it to be intelligent; indeed, much more so than the average human being.

This, of course, is not meant as a scientific test for intelligence. It is a sociological test for intelligence, which is apt, because the word "intelligence" as it is normally used has more sociological meaning than scientific. In particular, we say that someone is "very bright" if they seem able to jump through all the intellectual hoops which we've seen them encounter in everyday life.

> I'll be spending many hours on airplane flights over the next three
> weeks or so: Anyone with recommended reading on these topics, speak
> now. (_Goedel, Escher, Bach: An Eternal Golden Braid_ need not
> apply. Have it. Read it. Likewise, Penrose's _The Emperor's New
> Mind_: Have it, partway through it.)

I've generally been very disappointed in books I've read that are specifically about artificial intelligence. This includes Penrose's first book, which was terrific when he wasn't talking about intelligence, and pretty bad when he was. "Goedel, Escher, Bach" was pretty good.

Only two books spring to mind as being especially worthwhile reads among the books I've read on the subject:

Minsky's "Society of Mind" (1985) had a considerable influence on the way my thinking has developed. However, I read it as an impressionable, none-to-critical twelve year old, so I'm not sure how it has dated. Definitely worth a look.

Ray Kurzweil's "The Age of Intelligent Machines" (1990?) was a good read. A little dated now, it still contains a considerable amount of useful information, and some thought-provoking points.

I've "read", to varying degrees, a couple of dozen other books which are either directly about AI, or which touch on it as a major theme. Even the worst manage to stimulate some thoughts, but I recommend shorter essays or sf novels as being just as stimulating. Daniel Dennett seems to be entertaining and thought-provoking, although I haven't read much of his material.

> I envision my proposed "Extended Answers" as essay style answers, for
> what it's worth.

That would be useful. One of the advantages of (two-way) hypertext is that these essays could stand on their own, while still being an integral part of a >H FAQ.

Michael Nielsen