The current pace of peer review just isn’t compatible with the rate of improvement in AI models. It’s an exercise in absurdity for reviewers (who aren’t paid btw) to spend months on papers that will likely be far superseded by the time their reports are complete.