Re:Artificial General Intelligence: Now Is the Time
by jey on Tuesday June 05, 2007 @12:10PM
(#26470)
I'm suggesting that we fund research into the problem of AGI. I'm not saying that we should be funding teams to code up someone's personal AGI hypothesis. In the past, too much attention has been focused on trying to code up an AI without any real justification for why that AI system would work, and this has led to the failures you cite. What we need is fundamental research into intelligence, theories that describe intelligence and how to attain it. This is how research progresses in other fields; rocket scientists do not come up with their personal favorite hypothesis of rocket propulsion then seek funding to try to build a rocket using that propulsion system. Instead, they form models and consistent theories to validate their hypotheses before actually attempting to build a rocket. Why should AI research be any different? If the implicit assumption is that general intelligence is fundamentally just too difficult to understand, I disagree, and have not seen a convincing argument making that case.
Re:Artificial General Intelligence: Now Is the Time
by ced on Wednesday June 06, 2007 @06:26AM
(#26576)
Theories are ultimately validated through experimentation and observation. For "intelligence", that means coding it up, and see if it learns anything. That doesn't sound very different to me from "funding teams to code up someone's personal AGI hypothesis".
I'm sure that a lot of researchers in academia have pet theories about how to make it happen, but they need grants. The NSF has funded enough failed AI projects, it's hard for them to justify funding more given the current lack of money in science.
There are probably a lot of people working on AGI on their own, or in startups. I am. But you won't hear much about people like me unless we succeed to some level.
IMO, the field is waiting for its Einstein to come up.