ICML Keynote on Automated Science
math_rachel · x · 2026-07-15
This post shares the slides and full transcript of an invited ICML 2026 keynote titled "What will be left for us to work on?", highly recommending a full read.
A core takeaway is that viewing "automated science" merely as problem-solving is a fundamental misunderstanding. The goal of science isn't just to get tasks done; it's to form human understanding. The author emphasizes that human understanding is not friction to be eliminated by automation—it is an indispensable part of science itself. Without it, we lose all the derivative value it brings.
The keynote also distinguishes between "recursive self-improvement, human-like AI, superintelligence, and economically transformative AI," arguing that these concepts do not inherently imply one another.
More from AGI Musings
- BCI researcher: augmentation won't guarantee immortality, but beats the alternative — danfaggella · 2026-10-27
- OpenAI releases Signals dataset comparing ChatGPT work usage across nearly 100 countries — soumitrashukla9 · 2026-10-27
- Airbnb CEO Chesky: pure people managers and change-resisters won't survive the AI era — fortune · 2026-10-27
- Reddit question: If AI creates universal abundance, why invest at all? — AlephMartian · 2026-10-27
- Researcher: assume all LLMs are situationally malicious — never run agents outside a VM — jd_pressman · 2026-10-27
- Delta, Chase and Amazon are blocking AI agents — and the blockers may lose, predicts 6-stage war — alvelda · 2026-10-27