Comments on: Watson: supercharged search engine or prototype robot overlord?

By: whereIsHal

whereIsHal — Fri, 11 Mar 2011 22:40:28 +0000

Having watched the hour long Nova (PBS) program on the development of Watson, I was fascinated by the fact that Watson didn’t make a lot of progress in competing with humans merely by looking up answers in a database. What made all the difference was when they incorporated machine learning (which doesn’t seem to show up in Wolfram’s diagram). I took the machine learning to mean a sophisticated neural network that used pattern recognition.

By: martoom

martoom — Thu, 10 Mar 2011 11:07:16 +0000

The problem with these types of mind simulations is that rules are imposed from the outside. The more flexible the program is meant to be, the more algorithms are necessary. The comes a time when richness gives way to internal incoherence.

The answer lies in a system that runs one algorithm which allows connections to be made according to cognitive/semantic affinity, and its representative states having such a granularity that any ‘rule’ can be recreated, provided the input provides the appropriate context in the first place. Like the human mind.

Taking care of simulating those functionalities and there is no reason why a human-like mind cannot have an artificial origin.

By: jstjnk

jstjnk — Fri, 18 Feb 2011 22:34:03 +0000

Hmmm, so the Human intelligence “Ben G.” is so stuck in his limited social domain he’s unable to distinguish what a non-geek looks like. :-) hehehehe

Perhaps Humans aren’t quite as “domain free” in their thinking as some like to think they are.

Justin

By: technoguyrob

technoguyrob — Fri, 18 Feb 2011 06:09:47 +0000

“But … well … even so, it was pretty damn funky watching an audience full of normal-looking, non-AI-geek people sitting there cheering for the victorious intellectual accomplishments of a computer!”

Actually, the Jeopardy audience that night was composed entirely of IBM employees (engineers, staff, and even some executives).