From "LLMs can't reason" to cracking Millennium Problems: the goalposts keep moving on AI consciousness

ctjlewis · x · 2026-09-28

ctjlewis traces the shifting AI reasoning debate: saying models "reason" was once controversial, when they failed at rhyming and basic React apps. Now that models crack Millennium Problems, critics say it's "not really reasoning" or that consciousness remains off the table — with almost no philosophical discussion of what the math breakthroughs mean.

Related event: Developer ctjlewis reaffirms long-held stance on machine consciousness in thread series(8 posts)→

Original post →

More from AGI Musings

AGI Musings channel →