AI Radar

Noteworthy AI conversations and writing, summarized from publisher notes.

OpenAI's Nick Turley on ChatGPT's Path to a Super Assistant

· BG2Pod with Brad Gerstner and Bill Gurley · BG2Pod

Nick Turley, who leads ChatGPT at OpenAI, joined BG2Pod to discuss how the product scaled from a one-month demo intended to be wound down to roughly 900 million weekly active users in three and a half years. He walked through ChatGPT's growth framework—what he attributes roughly one-third each to classic friction removal (like dropping the login wall), core product investments (search, personalization, writing blocks done jointly with research), and underlying model improvements including the 5.3 and 5.4 updates. Turley framed the next phase as building a "super assistant" that combines two compounding capabilities: action-taking agents and proactivity. He argued that general-purpose agents are close to escape velocity, pointing to domain successes like Codex in code, and that proactive surfaces such as Pulse will become far more useful once they can connect to user data and take actions on the user's behalf. He also discussed pricing as something that "has to change" given that unlimited plans don't make economic sense when test-time compute lets intelligence scale arbitrarily, and positioned OpenAI's ad pilots and partnerships (including Apple and Reliance) as ways to broaden access globally. On internal operations, Turley described a recent "Code Red" that paused multiple projects to focus the company on reliability, latency, and the basics of the ChatGPT experience, culminating in the GPT-5.3 and 5.4 releases. He discussed the challenge of allocating scarce GPUs across consumer, Codex, and research workloads, calling it the most painful zero-sum tradeoff. He also confirmed the hire of Peter Steinberger from OpenClaw and hinted at more to come. The conversation closed with rapid-fire questions where he praised NotebookLM, advised students to cultivate curiosity, and named AI-native professional services firms as the startups he's most bullish on.

AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.