[LINK] Talking about AI

[David Lochrin]

On 2016-02-12 11:32 JanW  wrote:

> Do any linkers remember back in the 70s that there was a competition between AI research and another similar angle? I'm at a loss what it was, but it was the more reasonable development in that conceptual area. It was before machine learning as a serious topic, too. Help!

I remember confident promises in the early sixties that systems with "artificial intelligence" would very quickly become as intelligent as humans (remember HAL?), but that glorius day kept receding.  Eventually "rules based" systems proved much more promising in terms of actual usefulness but they were manifestly not intelligent.

Dear old DEC ran an AI program.  I believe the original impetus came from a large British pie manufacturer who had an employee with an uncanny flair for predicting the number of pies to bake each day.  He'd sniff the air, consider the football matches to be played, consult the weather forecast & his own intuitions, and he was almost always right.  But this genius was way past retiring age so the baker consulted DEC, and they proposed to incorporate the forecaster's knowledge in a rules-based software package.  I'm afraid I don't know how successful it was.

One interesting question about the idea of "artificial intelligence" is what the expression might actually mean.

Consider the problem of perception.  The organic brain is a neural network which could, in principle, be copied in solid-state technology with cameras providing a close analogue of eyes.  When this device was switched on we would expect it to behave according to the circuit design, nothing more or less.  But could we expect it to perceive a "red" fire engine, a "blue" sky, and "green" grass?  There is no colour in physics, only electromagnetic radiation of varying wavelengths (the classical model) or photons of varying energy (the quantum model).


Aploogies for the delay in replying, I've been moving house.

David L.