The numbers there seem odd (or did before the page broke, possibly due to the HN 'hug of death')
90% chance of near term "human level" AI but only a 75% chance someone tries to influence the US election with "AI driven misinformation" which for most definitions of "AI" and "misinformation" is already happening?
Hug of death is fixed, even this modest traffic was too much for our cache.
| 90% chance of near term "human level" AI but only a 75% chance someone tries to influence the US election with "AI driven misinformation" which for most definitions of "AI" and "misinformation" is already happening?
Yes, if 2040 is "near term". The majority view is that before 2025, the only major AI safety concerns are deepfakes, misinformation, and maybe social engineering.
But over the subsequent years, the consensus shifts to predicting widespread disruption.
90% chance of near term "human level" AI but only a 75% chance someone tries to influence the US election with "AI driven misinformation" which for most definitions of "AI" and "misinformation" is already happening?