zlacker

Some documents on AM and EURISKO

submitted by sctb+(OP) on 2018-11-13 18:49:01 | 32 points 10 comments
[view article] [source] [go to bottom]

NOTE: showing posts with links only show all posts
2. erikj+PY3[view] [source] 2018-11-15 13:03:26
>>sctb+(OP)
There are more documents in this archive: http://nosuchlabs.com/pub/lenat-leaks.tar.gz
◧◩
5. khafra+M64[view] [source] [discussion] 2018-11-15 14:23:51
>>panic+i24
It's not just cute, it's an open problem in the theory of recursively self-modifying AI design--there's no general solution to an AI hijacking its own reward channel, so far. https://www.lesswrong.com/posts/upLot6eG8cbXdKiFS/reward-fun... is the most recent thing I know of in the area.
◧◩
7. inetse+m94[view] [source] [discussion] 2018-11-15 14:45:32
>>ASipos+kX3
There is a link to a basic AM clone, "Pythagoras" by Steven L. Tanimoto. http://web.cecs.pdx.edu/~mperkows/CLASS_ROBOTICS/LISP/tanimo...
◧◩
8. TezlaK+xH5[view] [source] [discussion] 2018-11-16 01:44:49
>>Cybiot+i94
There is no credible documentation of EURISKO's achievements. No one has every actually seen EURISKO's source code except for Lenat.

Despite huge advances in technology, in computational power, and in the amount of engineers and grad students thrown at these problems, the magical reasoning feats have never been reproduced by anyone else - not even by Lenat's much more expensive project Cyc.

What's more, Lenat did not give a consistent account of the Traveler tournament. One time he claimed "ninety-six ships in Eurisko’s fleet, most of which were slow and clumsy because of their heavy armor", other times he would claim the winning strategy was "astronomical number of small ships like P.T. boats, with powerful weapons but absolutely no defense and no mobility".

The most probable explanation is that there's more myth than truth in the stories about EURISKO.

(on a less serious note, the jargon file entry for "bogosity" references Lenat; see http://itre.cis.upenn.edu/~myl/languagelog/archives/003515.h...)

[go to top]