On Intelligence and Searle's Mind

Posted 6 May 2005 at 17:21 UTC by steve

The Swirling Brain sent a link to a recent Reason Online review of Jeff Hawkins' new book, On Intelligence and John Searle's book Mind: A Brief Introduction. Both authors fall into the camp that seem to still believe there is some sort of magical, non-deterministic something that makes meat brains better than machine brains. Hawkins thinks he has some ideas for making machine brains better, perhaps good enough to compete with meat brains. Searle revisits yet again his Chinese Room "thought experiment" which proves (he thinks) that AI is impossible. In effort to avoid the "dualism" label, Searle has invented the term "biological naturalism" to describe his philosopical position. The article mentions Daniel Dennett as well. Dennet's books Consciousness Explained and Freedom Evolves provide a sensible alternative for explaining consciousness and free will in a deterministic universe without excluding the possibility of machine intelligence.

Out of his box, posted 6 May 2005 at 18:51 UTC by dogsbody_d » (Master)

Searle's Chinese Box argument is akin to Zeno's paradoxes. It's just so darned clever. You know it's wrong, but explaining why can lead to the sort of difficulties that you usually only get with the offside rule. (That's in football by the way).

Chinese room, posted 6 May 2005 at 23:31 UTC by motters » (Master)

The Chinese room has always baffled me. It's sometimes used as an argument against strong AI. But wait, aren't we all in the Chinese room anyway? The biological neurons which constitute our brains communicate via elecro-chemical signalling. Individually each neuron has absolutely no understanding of the english language (or Chinese), it is simply following a mechanistic set of rules governed by genetics and the laws of chemistry and physics.

People who use the Chinese room to argue against AI are really also denying their own intelligence.

What do you think philosophers do?, posted 8 May 2005 at 02:58 UTC by dogsbody_d » (Master)

What you have yerself their sonny is "The Systems Reply."

I suggest you type "Chinese Room" into google and visit the first link. Oh go on then, I'll treat you http://www.iep.utm.edu/c/chineser.htm

Or read this, paper fans.

Searle, John. 1980a. "Minds, Brains, and Programs." Behavioral and Brain Sciences 3, 417-424.

Chinese room dosent consider simple heuristic AI, posted 9 May 2005 at 12:49 UTC by c6jones720 » (Master)

I dont think the chinese room analogy works when you consider a digital system that learns I/O relationships based on feedback from the outside world. This is heuristic AI at its simplest.

My understanding of the Chinese room is of symbol manipulation based on a lookup table (i.e a program). If there is no program in the first place it is still possible for simple digital circuits to form and learn I/O relationships based on feedback.

Aspect of simple AI are possible and have been possible for a long time. Maybe one of our stumbling blocks is we're trying too hard perhaps we should use lots of simple methods together.

The flaw in the argument, posted 9 May 2005 at 14:18 UTC by devlin » (Journeyer)

I consider the Chinese Room Argument, like many objections to AI, to be hoisted by its own canard. The key flaw to me is the assumption about the rule book, which seems to trivialize intelligent responses to Chinese queries. Either NLP is trivial (which AI opponents would argue against in another breath) in which case human intelligence is nothing special, OR that rule book is so complex and capable as to represent a legitimate intelligent schema. They (AI objectors) can't have it both ways.

BTW, in reading Hawkin's book I didn't get the impression that he considered human intelligence magical--rather the opposite--but he rightly acknowledges that there is a lot we haven't figured out.

See more of the latest robot news!