Gary Marcus Pushes Back on Antitrust Suit's 'Slowing AI Harms Subscribers' Claim
GaryMarcus · x · 2026-09-20
Gary Marcus weighs in on an AI antitrust complaint. On page 5, the plaintiffs argue that an agreement to slow model improvement lowers the quality subscribers get for the price they pay, amounting to an overcharge of the kind antitrust law was meant to prevent.
Marcus says he skimmed the whole filing and can sort of see the argument if he squints, but believes the real goal is to use antitrust law in a way other than originally intended — to stop companies from slowing down for safety reasons and to extract damages if they don't.
He counters that value to subscribers would only drop if an existing model were nerfed, and that companies shouldn't be obligated to keep making better models — just as no airline is obliged to replace the Concorde. He says he may write about the antitrust issues in October.
More from AGI Musings
- Software Is Receding: Personal Task Agents Become the Wedge for the Agent Platform Race — manosaie · 2026-09-20
- Scoble defends Waymo: humans kill 40,000 a year driving in the US alone — Scobleizer · 2026-09-20
- Model solving Navier-Stokes burned tokens equal to 4,000 years of human thinking — Dr_Singularity · 2026-09-20
- 45 expert scientists stress-test AI reviewers against Nature-family peer reviews — windx0303 · 2026-09-20
- Anthropomorphizing AI is fine when trillions are spent making it human-like — AndyMasley · 2026-09-20
- Engineer argues AI fingerprints in code and text are just lower-quality output, harming human communication — ryunuck · 2026-09-20