Nick Turley
Head of ChatGPT at OpenAI. Joined OpenAI when it was still a research lab, started without a defined role (first task: fix the blinds), and organically became the product leader for ChatGPT — taking it from a hackathon prototype to 700 million weekly active users. First major podcast appearance on Lenny’s Podcast, recorded the week GPT-5 launched.
Background: philosophy and computer science (wrote senior thesis on whether rational people can disagree); product leader at Dropbox (real-world product); product leader at Instacart (learned to operate at crisis pace during pandemic); jazz pianist (competition-level, nearly went to music school). Joined OpenAI after getting nerd-sniped while trying to get off the DALL·E waitlist.
Talks and interviews
| Source | Date |
|---|---|
| Nick Turley on ChatGPT | 2025 |
Recurring themes
- Empiricism over planning — with AI products, properties are emergent; you cannot reason about use cases in advance; ship to learn.
- “Is it maximally accelerated?” — forcing function for cutting through blockers and identifying critical path.
- The model is the product — no meaningful distinction; iterate on the model using product development discipline.
- Run towards high-stakes use cases — don’t disable medical/relationship use cases; build them well.
- Natural language yes; chat paradigm no — “ChatGPT feels like MS-DOS.”
- Curiosity as the key attribute — for non-research hires, more predictive of success than prior AI experience.
Key concepts attributed to Nick
- Sycophancy — structural analysis and measurement framework
- Evals — “lingua franca” framing for product–research communication
See also
- Nick Turley on ChatGPT
- Boris Cherny — parallel AI product leader at Anthropic
- Cat Wu — parallel Anthropic PM
- Evals
- Sycophancy