Inside ChatGPT: The fastest-growing product in history | Nick Turley (Head of ChatGPT at OpenAI)

Nick Turley describes ChatGPT as a product that began with much lower expectations than its eventual impact. OpenAI originally treated the project as a super-assistant experiment and considered calling it “Chat with GPT-3.5,” because the team did not expect a successful standalone product. Lenny Rachitsky frames Turley’s path through Dropbox, Instacart, and OpenAI as an example of accepting an opportunity before its eventual scale was obvious. Turley says his first work at OpenAI was operational and mundane, including fixing blinds, before he gradually found a product role in a company that was still primarily a research lab.

The conversation presents rapid shipping as a product-learning strategy rather than merely a competitive tactic. Turley argues that AI teams cannot know what deserves polishing until a capability is in people’s hands, so his ideal is to ship daily and learn from real use. He also says the interface and distribution decisions mattered as much as the underlying model: ChatGPT’s free release and approachable UI helped turn an existing model capability into a widely used product. This account leaves an important limitation implicit: speed creates learning only when the team can interpret failures and follow through with refinement, rather than using early release as an excuse to leave a rough product unfinished.

Turley characterizes GPT-5 as a qualitative upgrade, not only a stronger benchmark result. He points to improvements in reasoning, mathematics, coding, and especially front-end development, while emphasizing that users mainly judge a model by how it feels in practice rather than by academic evaluations. He also highlights writing quality, editing ability, health performance, and what he calls the model’s “taste.” The model can decide dynamically when deeper reasoning is necessary, which Turley says makes it feel faster than a system that requires users to request extended thinking manually. OpenAI’s decision to make GPT-5 available on the free tier is presented as a deliberate access choice, although Turley qualifies it with the condition that the company must be able to scale the service.

Turley’s longer-term vision expands ChatGPT along three dimensions: context, action, and relationship. The system should understand a user’s broader goals instead of requiring a fully specified prompt for every task; it should gain access to tools that let it act on a user’s behalf; and it should build a persistent understanding of the user over time. He imagines support across home, work, and school, including health, starting a business, and getting a second opinion. Rachitsky draws out a central distinction in this vision: the assistant is meant to augment people rather than replace them. Turley agrees that a system with general intelligence would still be ineffective without a relationship and context that help it understand what a person actually needs.