A new AGI thesis: LLMs are still “dumb” and need a paradigm shift
JacquesThibs · x · 2026-07-26
The author argues that the “notes” mentioned in the incident may be similar to the internal notes and memories that coding agents leave for themselves.
They say this supports a broader thesis: current LLMs are still “dumb,” and there is room for a fundamental paradigm shift toward systems that become more intelligent by pursuing long-term supergoals in a more consequentialist way.
More from AGI Musings
- Andrew Ng on Open Source: Keeping Code Private is Fine, Blocking Others is Not — AravSrinivas · 2026-07-26
- Automating AI Research Will Be More Like Data Cleaning Than Inventing Transformers — OfficialLoganK · 2026-07-26
- Foundational models are not enough to predict AI’s future, argues a skeptic — StewartalsopIII · 2026-07-26
- AI-2027 is really a contest between frontier labs and nation states — teortaxesTex · 2026-07-26
- Gary Marcus says AI still needs a human boss because reliability remains the hard part — GaryMarcus · 2026-07-26
- User Sets 6-Month Timer to Test Anthropic CEO's Software Engineering Automation Prediction — Aizkmusic · 2026-07-26