Eliezer S. Yudkowsky, <sentience@pobox.com>, writes: > Which is more important: The next level up in programming languages, or > a EURISKO-like reusable AI core?
This sounds like Cyc (www.cyc.com), but that is looking more and more like a failed project.
Here is an interesting excerpt from the author:
: In the late 1970s I built a computer program (Eurisko) that discovered
: things on its own in many fields. To get it to work, I had to give it the
: power to tinker with its own learning heuristics and its own goals. I
: would leave it running overnight and hurry in the next morning to see
: what it had come up with. Often I'd find it in a mode best described as
: "dead." Sometime during the night, Eurisko would decide that the best
: thing to do was to commit suicide and shut itself off. More precisely,
: it modified its own judgmental rules in a way that valued "making no
: errors at all" as highly as "making productive new discoveries." As soon
: as Eurisko did this, it found it could successfully meet its new goal
: by doing nothing at all for the rest of the night. This reminds me of
: HAL's boast: "No 9000 computer has ever made a mistake." I eventually
: had to add a new heuristic to Eurisko-one it couldn't modify in any
: way-to explicitly forbid this sort of suicide.
It's too bad that Lenat's brilliance has been sucked into the abyss of Cyc. If he had continued to work on Eurisko, perhaps with the goal of creating a system which could learn from experience the way babies do, he might have built a system which could develop common sense for itself rather than having to have it spoon-fed.
Hal