DeepMind essay: AI is evolving from research tool into an 'invention of a method of invention'
JMateosGarcia · x · 2026-10-08
Google DeepMind's James Manyika and AGI Economics director Alex Imas published a long-form essay, "Bending the Curve of Discovery," on AI's role in science.
Key points
- Research productivity faces headwinds: ideas are "getting harder to find" in semiconductors, agriculture, and pharma, with exponentially growing teams and capital needed; scientists face deepening specialization and rising coordination costs.
- AI's real promise is not accelerating existing practices but acting as a Griliches-style "invention of a method of invention" (IMI)—like the microscope or statistical inference—unlocking questions beyond human reach.
- Today LLMs and specialized models like AlphaFold act as economic complements, but handoffs still run through the scientist.
- The future: LLMs orchestrating those handoffs automatically—prompting specialized models, auditing outputs, and iterating until questions are answered.
The essay maps how scientists use AI today and charts AI's evolution from accelerating tool to an engine expanding the frontiers of knowledge.
Related event: DeepMind Reviews AI for Science: From Tool to "Invention of Invention"(3 posts)→
More from AGI Musings
- AI plus formal tools compress years of research into hours as community live-tracks integer multiplication breakthrough — aran_nayebi · 2026-10-09
- Mustafa Suleyman reacts with a single melting-face emoji to Anthropic's 'no Claude abuse' policy — mustafasuleyman · 2026-10-09
- David Manheim: open models shift AI power not to users but to NVIDIA — davidmanheim · 2026-10-09
- ARK Analyst: Superintelligence Will Be History's Best Energy-to-GDP Converter — DMaguireARK · 2026-10-09
- Debating FDT's decade of stalled progress: ex ante optimality as an alternative frame — jessi_cata · 2026-10-09
- Nathan Lambert to argue open science can still tackle AI safety and diffusion — natolambert · 2026-10-09