unhobbling
Leopold Aschenbrenner's term for removing artificial constraints on AI models (lack of tools, memory, context, reasoning time) to unlock latent capability without fundamentally new intelligence.
As defined in AGI is still 30 years away — Ege Erdil & Tamay Besiroglu
Where it comes up · 4
- Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history4 claims
- AGI is still 30 years away — Ege Erdil & Tamay Besiroglu1 claim
- Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history1 claim
- Dario Amodei1 claim
Most important claims · 7 of 7Search all claims for “unhobbling” →
- AI labs should maintain state-level security comparable to nuclear weapons programs because algorithmic breakthroughs, training techniques (like RL-based unhobbling), and model weights can all be stolen relatively easily—current security is at Level 0 (vulnerable to state activity) when it should be at Level 4, and a single person recently stole valuable AI code by copying it to Apple Notes and exporting as PDF, bypassing monitoring.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- RLHF (reinforcement learning from human feedback) unhobbled GPT-2's raw capabilities into a functional chatbot, achieving roughly a 100x improvement in human preference ratings compared to non-RLHF models, and enabled basic chain-of-thought reasoning, but the underlying model still possessed many unutilized raw capabilities that could be further unlocked.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- By 2025-2026, AI models will be basically smarter than most college graduates, but their economic usefulness will depend heavily on 'unhobbling' — transitioning them from limited chatbots to agentic systems capable of long-horizon tasks.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- GPT wrapper companies are making a poor bet because they're betting on stagnation—assuming intermediate models will require extensive schlep to integrate—when actually the trajectory points to drop-in remote worker AGI that will make wrapper functionality obsolete, sonic boom-ing past the intermediate capability regime.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- RLHF does not make a model smarter or merely appear smarter; it bridges the gap between a smart model and the human — like a very smart person who can't communicate — so it 'unhobbles' the model. The kind of RLHF done today mostly doesn't add raw intelligence, though other RL can.Dario Amodei
- Much of AI's economic usefulness depends not on raw intelligence but on 'unhobbling'—unlocking capabilities like long-horizon agentic tasks and computer use; by 2027-2028, unhobbling will transform models from chatbots into drop-in remote workers that you interact with like coworkers (Zoom, Slack, assigning projects), which is far easier to integrate than intermediate models that require extensive workflow 'schlep'.Leopold Aschenbrenner — 2027 AGI, China/US super-intelligence race, & the return of history
- ChatGPT exemplifies cheap 'unhobbling': roughly 1% of additional compute spent on post-training a pretrained internet corpus was enough to make a model highly competent at chatbot conversation, and reasoning similarly emerged from a small fraction of compute spent on RL.AGI is still 30 years away — Ege Erdil & Tamay Besiroglu