Gary Marcus Argues Generative AI Has Failed to Deliver on Promises
Published 2026-02-25Foundation ModelsMedium
Summary
Gary Marcus, a prominent AI skeptic and cognitive scientist, published a post on his Substack titled "Turns out Generative AI was a scam," arguing that generative AI is "very very far from what it has been cracked up to be." The piece appears to consolidate his longstanding critiques of large language models, questioning whether the technology has delivered meaningful value relative to the enormous investment it has attracted. The post generated significant discussion in the comments, with part
Alignment: Challenges current position
generative-ai-criticismllm-limitationsai-hype-cycleai-assisted-codinggary-marcusai-roifoundation-model-skepticismsoftware-engineeringai-code-quality