Steven Johnson is a very meta writer. He writes frequently about science and technology, and likes to immerse himself in what he covers, even using it to change the way he writes his books. A few years ago, a few months before ChatGPT launched and the AI craze took the tech world by storm, Johnson was commissioned by a magazine to fall into the deep hole of AI, and never came back.
In addition to writing books, Johnson also currently works at Google. He is part of a team developing a product called NotebookLM, which the team calls “Notebook.” It’s a note-taking and research tool where you upload documents and import web links, and Notebook’s Gemini-powered AI helps you organize things, extract information, and deepen your understanding of the subject. “Google reached out to me and said, ‘You’ve always dreamed of the ideal software tool to help you organize your thoughts, write sentences, form connections, and brainstorm. Now we can make it happen.'” Johnson signed a contract and has been working at Google since the summer of 2022.
The product itself was first released as Project Tailwind in 2023, and has since been rebranded and significantly expanded. Last week, the team released Audio Overviews, which generates a podcast based on information provided by users, complete with two chatty hosts, lots of back-and-forth, and a truly surprising attention to phrases like “deep dive” and “fasten your seatbelt.” It’s fascinating, complex, and rapidly improving.
In this episode of The Vergecast, Johnson talks about his interest in AI, his experiences at Google, and the present and future of NotebookLM. We discuss the complex problems these tools raise and whether it’s OK to entrust AI to do your research and homework.
We also talk about how tools like NotebookLM make fact-checking accurate and easy, why the context window is more important to the future of AI than many realize, and how often AI podcast hosts say “like” to a conversation. We also talk about Johnson’s own process as a writer and creator, and how AI is changing the way he works.
If you want to learn more about what we discussed in this episode, start with these links:
