Not superintelligence but OpenAI and Anthropic's agent business models is the real danger, argues clairlemon

GaryMarcus · x · 2026-09-24

Gary Marcus amplified a take from clairlemon arguing that the real danger isn't 'superintelligence' but the business models of OpenAI and Anthropic. The companies push enterprises toward autonomous agents because agents consume far more tokens — yet 'agency' isn't 'intelligence,' and today's agents can be very stupid.

Original post →

More from AGI Musings

AGI Musings channel →