The case against ai “fast takeoff”
https://www.interconnects.ai/p/lossy-self-improvement
Unpredictable chaos develops over years, not weeks. That’s the optimistic case now.
The case against ai “fast takeoff”
https://www.interconnects.ai/p/lossy-self-improvement
Unpredictable chaos develops over years, not weeks. That’s the optimistic case now.
@jgordon that was a good summary of competing factors. The original concept of The Singularity hinges on superhuman intelligence. Now we face the possibility of a lesser singularity supported by lesser intelligences (human and machine).
I linked earlier to an options trading experiment in which voting by fallible AIs made possible a "don't know" output. Not only that, a "do nothing" signal.
Things like that might allow progress without excess risk.