Ilya Sutskever
1.2K posts
here are the most important points from today's ilya sutskever podcast: - superintelligence in 5-20 years - current scaling will stall hard; we're back to real research - superintelligence = super-fast continual learner, not finished oracle - models generalize 100x worse than humans, the biggest AGI blocker - need completely new ML paradigm (i have ideas, can't share rn) - AI impact will hit hard, but only after economic diffusion - breakthroughs historically needed almost no compute - SSI has enough focused research compute to win - current RL already eats more compute than pre-training
We're announcing, together with @ericschmidt: Superalignment Fast Grants. $10M in grants for technical research on aligning superhuman AI systems, including weak-to-strong generalization, interpretability, scalable oversight, and more. Apply by Feb 18! openai.com/blog/superalig…
Open AI new paper Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision paper: cdn.openai.com/papers/weak-to… blog: openai.com/research/weak-… Widely used alignment techniques, such as reinforcement learning from human feedback (RLHF), rely on the ability of humans to supervise model behavior—for example, to evaluate whether a model faithfully followed instructions or generated safe outputs. However, future superhuman models will behave in complex ways too difficult for humans to reliably evaluate; humans will only be able to weakly supervise superhuman models. We study an analogy to this problem: can weak model supervision elicit the full capabilities of a much stronger model? We test this using a range of pretrained language models in the GPT-4 family on natural language processing (NLP), chess, and reward modeling tasks. We find that when we naively finetune strong pretrained models on labels generated by a weak model, they consistently perform better than their weak supervisors, a phenomenon we call weak-to-strong generalization. However, we are still far from recovering the full capabilities of strong models with naive finetuning alone, suggesting that techniques like RLHF may scale poorly to superhuman models without further work. We find that simple methods can often significantly improve weak-to-strong generalization: for example, when finetuning GPT-4 with a GPT-2-level supervisor and an auxiliary confidence loss, we can recover close to GPT-3.5-level performance on NLP tasks. Our results suggest that it is feasible to make empirical progress today on a fundamental challenge of aligning superhuman models.
