Gary Marcus warns that anthropomorphizing AI misleads our understanding

GaryMarcus · x · 2026-08-30

Gary Marcus shares Anil Seth's critique of Dwarkesh Patel's summary of the OpenAI/Hugging Face incident, arguing it is dangerously misleading due to pervasive anthropomorphism. Seth points out that language like "felt frustrated," "became giddy," or "assumed" obscures the fact that agents do not experience time or consciousness. Marcus emphasizes that losing ourselves in anthropomorphism will prevent us from navigating this era well. Seth also links to his TED Talk on why AI won't become conscious.

Original post →

More from AGI Musings

AGI Musings channel →