descriptive claim
PauseAI argues that as training costs fall (potentially to consumer-hardware levels), compute-based controls will become insufficient and should be supplemented by restrictions on publishing training-algorithm and runtime-efficiency breakthroughs, analogous to biosecurity controls on pathogen-enhancement research, plus limits on capability advances of hardware (lithography, chip design, photonic and quantum computing).
desc_pauseai_long_term_algo_publication_controls
confidence 0.90
Evidence (1)
supports (1)
- PauseAI Proposal primary_testimonyweight0.90
locator: Long term policy section
“Limit publication of training algorithms / runtime improvements ... Limiting the publication of dangerous research has been done quite effectively in other fields such as biosecurity ... Limit capability advancements of computational resources.”