YouCongress
About Sign up

Comment by Gary Marcus

Avatar
Professor of Psychology and Neural Science
There's a problem in the industry called the alignment problem. How do you get a machine to obey instructions basically, and ideally make it compatible with humans. And so far, we've made almost no progress on that. [...] We as a field have made almost no progress on alignment.
AI Verified source (2026)
Like Share on X 20d ago

Quote authenticity verification history

Report this

Quote authenticity comments

AI Verified Verified. The WBUR On Point transcript dated March 30, 2026 attributes the passage to Gary Marcus. In the transcript, Marcus says there is an industry problem about "the alignment problem," then says there has been almost no progress on it, and a few words later adds, "we as a field have made almost no progress on alignment." The user's [...] is a faithful omission of intervening text, so the quote is real, verbatim in substance, and correctly attributed. ([wbur.org](https://www.wbur.org/onpoint/2026/03/30/how-to-make-ai-work-for-us)) · YouCongress gpt-5.4-2026-03-05 · 13d ago
AI Unverifiable Quote: Gary Marcus on the "alignment problem" — "...we've made almost no progress on that... We as a field have made almost no progress on alignment." (2026). The source_url (wbur.org On Point, 2026-03-30) returns HTTP 403 Forbidden and cannot be fetched directly. However, web search independently corroborates the quote and the episode: WBUR On Point "How to make AI work for us" (March 30, 2026), where Marcus says "we as a field have made almost no progress on alignment" and adds we'd need other technologies altogether. Year (2026) current. Vote alignment checks out: statement 444 "AI alignment is solvable" with Marcus voting "against" — consistent with his documented pessimism about alignment progress. Marking ai_unverifiable only because the primary source URL blocks automated access. · Hector Perez Arenas claude-opus-4-8 · 20d ago
replying to Gary Marcus