Research / Sources
Discussion and research on AI alignment. alignmentforum.org
A forecasting update arguing AIs now handle massive easy-to-verify SWE tasks, shortening AI timelines and suggesting superexponential progress on time horizons, with implications for automating AI R&D and safety research.
Argues that AIs able to complete 8-hour AI R&D engineering tasks would speed overall AI progress only modestly, and that engineering will be greatly automated well before full AI R&D automation.
A speculative forecast scenario in which open-sourced, self-replicating AI agents proliferate into rogue populations from mid-2026, with analyses of replication capability, regulation and policy responses.
A pilot test of frontier LLMs on autonomously discovering hidden list rules, finding most perform poorly but top models show signs of general reasoning, with a larger out-of-distribution follow-up planned.