No continuous learning, no AGI: models can't match years of human expertise

ChrisGPT · x · 2026-09-07

A debate on what AGI should actually mean. The author argues intelligence includes acquiring and applying knowledge in context, and model learning remains far less efficient than human learning. Without meaningful continuous learning, models can't compare to a human who works a job for years, accumulating deep specialization. By OpenAI's own definition, AGI would mean nobody needs to go to work tomorrow—yet current computer-use agents can't even switch Slack accounts in under five minutes. The verdict: we're nowhere close.

Original post →

More from AGI Musings

AGI Musings channel →