Sunday 7 January 2018

Unconscious Artificial Intelligence

I have been misunderstanding something. Computers can be called "intelligent" if they play chess, invest money, control air traffic, fly drones, drive cars etc. I called such computer operations the simulation, not the duplication, of intelligence because I associated intelligence with consciousness, in particular with reflective self-consciousness. However, intelligent organisms are conscious whereas so far "intelligent" artifacts are not.

But science fictional AIs, e.g., in Poul Anderson's Harvest of Stars Tetralogy and Genesis, are conscious. Sometimes Anderson refers, inappropriately I think, to consciousness-level computers. At other times, he recognizes that a conscious artifact would have to be (something like) an artificial neural network, not a computer. Meanwhile, in the real world, artificial intelligence advances without becoming conscious. The sf identification of AI with consciousness, indeed with superhuman consciousness, seems to be a mistake.

2 comments:

S.M. Stirling said...

The same result can have different causes. Wheels and legs will both move mass, but legs are not wheels.

paulshackley2017@gmail.com said...

Kaor, Paul!

While I too am skeptical of "artificial intelligence" become CONSCIOUS and self aware, I enjoyed reading such speculations when done well, such as in Poul Anderson's HARVEST OF STARS books.

And the idea of computerized self driving cars really does interest me as a real advance in that rather boring, late 19th century technology (altho I still hanker for Andersonisan "air cars" seen in ENSIGN FLANDRY). Self driving cars means even blind persons could now have cars--the computer would drive the cars to where they wanted to go.

Sean