Gary Marcus showing up on the Dwarkesh Podcast right now would plug a leading AI skeptic into Silicon Valley’s favorite idea-lab just as the OpenAI vs Elon Musk lawsuit heads toward an April 2026 jury trial, raising urgent questions about governance, mission drift, and who gets to steer AGI. Marcus has been one of the clearest voices on why today’s LLMs hallucinate, why business models look shaky, and why society needs stronger oversight, and he has been proven prescient as the hype cycle wobbles.
In 2025 he called GPT-5 “underwhelming” and warned that a $2.7 trillion AI unicorn valuation pile is out of sync with revenues, a take that looks timelier by the week. Bringing him into Dwarkesh’s high signal audience, typically filled with AI builders, would force a real-time reality check on safety, data scraping, and economic sustainability.
As far as public episode listings show, Marcus has not appeared on this podcast before, which makes this a first-time conversation with fresh stakes for 2026. If you want sharper, more accountable AI debates in the room where tech listens, support this nomination today.
