Research
Paper argues System-1 screening could make AGI scaling more efficient
The paper proposes a predict-gate-execute-update loop where a lightweight predictor decides which candidate interactions System-2 runs. It claims execution compute falls by a factor approaching 1/ρu under ideal screening, with savings reduced by prediction errors, exploration and distribution shift.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead
Does Scaling Toward AGI Depend on What System-1 Lets System-2 Execute?