Very good question and which parties would be involved? Food for thought!
However, it appears the author of the following article meant something very different than the analogy to nuclear weapons agreements of the past century!
"... In fact, model quality is growing harder to compare as the industry questions the efficacy of standard benchmarks such as mathematical problem completion. These benchmarks can be gamed if the model is trained on the specific problems in advance. What’s more, as OpenAI researcher Noam Brown recently noted, the right definition of capability is not only what problems a model can solve but how quickly and at what cost. ..."
What would multilateral ‘AI arms control’ look like? (behind paywall)
No comments:
Post a Comment