LOGI Now Available in PDF Wednesday, Dec 2 2009
Levels of Organization in General Intelligence, a 2002 publication by Singularity Institute Research Fellow Eliezer Yudkowsky, can be considered our first attempt at an AI theory. (But as Eliezer points out in the comments, “LOGI still stands as an OK piece of work on the evolutionary psychology of human general intelligence. But as a theory of AI, it’s not even in the right league.”) The evolutionary psychology of general intelligence is profoundly understudied because it is much easier to look at narrow parts of the puzzle.
More recent work and thought can be found in the many hundreds of pages of writing in the Less Wrong sequences, which clarify the philosophical and epistemological foundation of our AI work. If you’re familiar with our work and are interested in working together with us, take note of our recent call for SIAI Visiting Fellows.
December 3rd, 2009 at 2:44 am
I can’t think of anything that is worth remembering about LOGI, FAI-wise (you talk about “draft of our AI theory”), beside the general sense that greater intelligence allows to do some impressive stuff. Cogsci-themed discussion seems askance to FAI, and part 3 is more applicable to people thinking of extending their cognitive capabilities than to actual AIs. Optimistic outlook on tinkering was also too strong.
To add another piece of opinion:
“LOGI’s out the window, of course, as anyone who’s read the arc of LW could very easily guess.” — EY.
December 3rd, 2009 at 11:56 am
LOGI still stands as an OK piece of work on the evolutionary psychology of human general intelligence. But as a theory of AI, it’s not even in the right league.
December 7th, 2009 at 3:18 pm
Clearly a replacement for LOGI is needed. No one will read hundreds of pages of often off topic LW entries to guess at the latest thinking.