Google researcher ibab: models trained on humanity's commons should be open-sourced
ibab · x · 2026-11-20
Shawn Presser (ibab), a core figure in the Gemma/open-source ecosystem, lays out an ethics argument in a long-form post:
- The problem: LLM pretraining distills humanity's collective work — including GitHub's open-source code — into weights. The content was released as a common good, yet AI companies turned it into hundred-billion-dollar businesses, selling coding models back to the very community they learned from.
- But strong copyright claims don't hold: like humans, LLMs learn and synthesize rather than reproduce; banning LLM learning would require explaining why humans are allowed to.
- The resolution: data from the commons → models back to the commons. Release open weights for anyone to download and use — the core idea behind open-weight models, which are starting to rival proprietary labs.
After years building LLMs, the author believes this will become the majority view and today's closed development will look odd in retrospect — "provided the models haven't taken over by then."
Related event: Google Researcher Argues Open Weights Should Return Distilled Human Commons(2 posts)→
More from AGI Musings
- Garry Tan backs Bostrom's 'if nobody builds it, everyone dies', urges taking Deep Utopia seriously — garrytan · 2026-11-20
- Superforecasters put 33% odds on a week-long outage hitting 1M+ people in an OECD country by 2028 — NathanpmYoung · 2026-11-20
- Sentinel brief 'Consider Cyberarmageddon' extrapolates cyberattack growth to societal failure — NathanpmYoung · 2026-11-20
- $1M into a public benchmark gets labs to spend $100M+ hillclimbing it — _jasonwei · 2026-11-20
- Stanford team at NeurIPS 2026: analogical reasoning fixes LLM mode collapse in science — james_y_zou · 2026-11-20
- Founder claims 'LLM arbitrage': spotted 17 counterparty errors in one day — StewartalsopIII · 2026-11-20