Experts predict emergence of Human-level AI
H+ magazine does an extensive survey attempting to nail down 21 experts on the specific timeline and impact of true artificial intelligence. Some predict Nobel Prize level scientific work within 30 years (for what that’s worth.) The Turing Test is mentioned. If you’re not familiar, that involves a kind of blindfold question-answer session, and for the machine to pass it would be required that the human would not be able to tell the respondent is artificial.
Far be it from me to question official experts but this is a the kind of thing that always seems to be 20 years away. The incentive of these experts to promote the possibility cannot be taken into account because then there’d be no story. The responses are interesting and as scary as they’re meant to be but no specifics on how harmful AI will manifest is forthcoming. Science fiction is invoked to provide the atmosphere.
Applying common sense, which always seems to befuddle experts, just the question of which approach to take is a give-away that reliance on such experts to find a clear path will get you lost on a hurry. Probability theory came out as the most popular approach? I don’t know why it was even a choice. I supposed there must really be AI researchers out there pursuing a pure probabilities solution to AI, but I’d think even the least intelligent human is getting it right well inside the margin of error, and the smartest of use are not walking around calculating the odds subconsciously.
For my money than I’d let it ride on one of those nonlinear dynamic systems.
AI, artificial intelligence | bshears February 11, 2010
