AI Researchers' Views on Automating AI R&D and Intelligence Explosions

A collection of recent AI policy, safety, and applied-research papers highlighting (1) substantial concerns that automating AI R&D could enable recursive improvement and rapid capability growth — a top risk cited by many frontier researchers — and the need for metrics to monitor AI R&D automation; (2) proposals for scalable, continuous assurance and governance of agentic systems (TAIP, AMAGF, Control Quality Score) to address loss-of-control, auditability, and operational-scale assurance shortfalls; (3) practical engineering work on trustworthy deployment (MLOps fairness gates, explainability,