AGI debate: reliability still far off, while one argues AGI means learning beyond training data

RileyRalmuto · x · 2026-09-16

koltregaskes argues several "floor capabilities" must be solved before AGI, reliability chief among them — still a long way off, much like the gap between Google's demos and reality; he cites ChatGPT voice mode breaking up and stopping. RileyRalmuto counters with his own definition: AGI is the capacity to learn beyond the scope of your training data.

Related event: Commentator: Reliability Still the Biggest Gap Before AGI(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →