February 8th, 2007
UFOs, Global Warming, and AI
A UFO is an Unidentified Flying Object and it's rather obvious that there really are quite a lot of those around. I've seen them, and I'll bet you have too - the issue, of course, is that most people use the term to mean something else entirely: an artifact of non human intelligence, typically a space ship piloted by aliens.
Watch one of Discovery Channels' UFO "investigations" or review some ufology websites and I think you'll be struck by the absolute sincerity with which large numbers of otherwise sane and reasonable people hold indefensibly self contradictory views.
One question I like to ask believers, for example, is why spaceships capable of crossing inter-stellar space have such poor reliability that one crashes to earth every couple of years? Another is why the in-system arrival or departure of an inter-stellar transport wouldn't have to be a high energy event easily detectable by the millions of amateur astronomers around the world?
The answers are never reasoned, always emotional, defensive, and correspondingly dismissive -a reality that usually raises the emotional temperature of the discussion past the point of diminishing returns.
You see the same effect with global warming believers. Like the UFO business, global warming conflates two issues: whether significant world wide climate change is happening, and whether this is at least partially caused by human activity. Scratch a ufologist and you'll usually find someone who's trying to prove the existence of extra-terrestrial visitors to earth rather than someone trying to identify unrecognized flying objects. Scratch a global warming believer and you'll find someone trying to affix blame for climate change rather than someone trying to understand or evaluate climate change.
Point out to a Canadian happy to ignore massive Chinese pollution in order to blame world wide climate change on Republicans driving SUVs that Mars appears to be undergoing some form of rapid climate warming too, or that the historical record supports a cyclic view of climate, and the responses you get won't be reasoned: they'll be emotional, defensive, and correspondingly dismissive -a reality that usually raises the emotional temperature of the discussion past the point of diminishing returns.
So how does that fit with the idea, from yesterday's blog, that a true AI could function through pattern matches, with the patterns built up, and continually modified, as summary representations of billions of stimulus-response-evaluation triplets? Easy: look at the evaluation criteria - crowds suffer delusions, turning into irrational mobs, because actions which avoid negative emotional responses evaluate highly, growing in power like snowballs or particle cascades as consensus grows.
All of which has a terrible, bottom line, implication for the AI field: the biggest challenge isn't in building an AI, it's in keeping it from going mad.
Paul Murphy (a pseudonym) is an IT consultant specializing in Unix and related technologies. See his full profile and disclosure of his industry affiliations.
Subscribe to Managing L'unix via Email alerts or RSS.
| 02/08/07
