Saturday, May 9, 2009

Making AI

I recently took Michael Anissimov to task over some of his fretful reaction to this topic.

Now, via Brian Wang comes notice of the US Army's DARPA-lead effort to achieve that development:

The program plan is organized around three interrelated task areas: (1) creating a theory (a mathematical formalism) and validating it in natural and engineered systems; (2) building the first human-engineered systems that display physical intelligence in the form of abiotic, self-organizing electronic and chemical systems; and (3) developing analytical tools to support the design and understanding of physically intelligent systems. If successful, the program would launch a revolution of understanding across many fields of human endeavor, demonstrate the first intelligence engineered from first principles, create new classes of electronic, computational, and chemical systems, and create tools to engineer intelligent systems that match the problem/environment in which they will exist.

If I understand this correctly (as ever, not a given), the .mil wants to identify and then engineer a non-human intelligence that could ultimately replace ... well, me. The Me I was low these more-than-one decades passed, the average soldier (sailor actually) who performs some necessary but less-than-glamorous role in accomplishing the Mission.

Leaving aside the potential ethical quandaries this might pose, I suggest that at least one of the issues DARPA wants to tackle as part of this project is that regarding the question of "friendly AI" that has Mr. Anissimov (among numerous others) properly concerned. One possible mechanism to arriving at such an objective must be to simply manufacture an entity that cannot continue to function so far outside the designed tolerance levels that designate "friendly". This would offer future military planners the ability to much more finely calibrate the requirements of any potential mission requirements as well as mute the costs of success.

Swinging the Sword of Damocles* does provide a different set of options than does sitting on it at least.

*I know, classically it hangs above one by a single horse hair. It works either way I suppose, but my version seems more true to the military ethos I think.

No comments: