Voices
Huang's actual position on pacing: keep labs in control, use many evaluators not one, and stop publishing p(doom) numbers
At the same All-In Summit appearance Huang said 'If you want AI to be safe, the first thing is we need to make sure that the labs that are building it are in control,' and endorsed independent evaluators with a caveat aimed squarely at METR: 'it's good to have independent auditors or evaluators, but they just have to have multiple.' On extinction estimates he was blunt: 'what does that mean, 10% of extinction? ... we shouldn't because it's made up,' calling the practice 'irresponsible.' He also argued open models matter for sovereignty, privacy and startups, and pointed at past AI forecasts that missed on radiology, code generation and near-term job losses.
Source
↳ Follow the thread