Ryan Greenblatt – What happens once AI can automate AI research?

kept by eddie aug 12

The discussion explores whether automating AI R&D could trigger recursive self-improvement, potentially yielding years of progress in months, with debate on verifiability, data bottlenecks, and alignment risks.

Rethinking the Data Moat
The piece contends that AI progress is driven chiefly by algorithmic advances and smarter data curation, not merely more data or human expert labels, citing Greenblatt and Bi.
also on Redwood Research, Ryan Greenblatt, #ai-research
What the hell happened with AGI timelines in 2025?
Expert AGI timelines contracted then expanded in 2025 as reasoning model gains proved costly to scale, though steady progress and growing revenue undercut radical pessimism.
also on #agi-timelines
🎙️Inside a Chinese AI Lab: How MiniMax Builds Open Models
MiniMax researcher Olive Song reveals how they debug RL training, handle model hacking, and prioritize alignment and engineering discipline in open-weight model development.
also on #alignment
9 theses on AI | Sarthak Munshi
AI progress is constrained by long-task reliability, labor reallocation, cost inefficiencies of general APIs, the declining value of raw coding skills, inadequate benchmark testing, the limits of formal verification without strong specs, memory-bound local hardware advantages, the shift from data to environment-driven training, and the rising competitiveness of US open-weight models.
also on Antithesis
engineer away the slop
The author announces joining Antithesis to promote formal verification and deterministic testing as essential tools for producing defect‑free software amid rising code volume.
also on Antithesis