Live with Tim O’Reilly

AI has led us to this amazing inflection point where all of the old habits of product development and product management need to be rethought. As Google’s Marily Nika predicts, “All product managers will be AI product managers in the future.” So we need to plan for it now.

On this episode of Live with Tim O’Reilly, Marily sat down with Tim to discuss how AI is fundamentally transforming product management, touching on emerging roles like AI engineers and agent engineers, the importance of human judgment in an AI-powered world, and practical workflows for rapid prototyping using AI tools like Perplexity, custom GPTs, and v0. They also explored the balance between leveraging AI’s capabilities and avoiding the “shiny object trap” of using AI for its own sake rather than solving real customer problems; discussed some tips for prospective PMs interesting in focusing on AI within their specific domains; and considered whether using AI is “cheating”—and why companies would do well to build more community around AI use.

For more highlights, check out Tim’s takeaways from the event on O’Reilly Radar.

AI companies are scraping content at unprecedented rates—with some platforms taking 1,500 pages for every visitor they send back, compared to Google’s historical 2-to-1 ratio—effectively breaking the basic value exchange that built the modern web. While this may be destroying the business model that has sustained online content creation for decades, it’s also turned the spotlight on some of the flaws inherent in the internet economy. As Cloudflare CEO Matthew Prince observed in this episode of Live with Tim O’Reilly, “A lot of what’s wrong with the world today comes back to the fact that Google taught us all that the thing that we were chasing was traffic.…But traffic is a bad approximation for value.”

Matthew and Tim thought through the question of value in a world where chasing traffic is no longer viable, discussing the urgent need for new compensation models that reward content creators both financially and through recognition, and imagining future scenarios ranging from a dystopian “Medici model” where a few AI giants employ all journalists to more decentralized approaches that preserve independent content creation. They also dove into Cloudflare’s innovative technical solution of blocking AI bots by default while requiring cryptographic identification—a first step in ensuring the creators who “fill the holes in the Swiss cheese” of human knowledge can still build sustainable careers and receive the recognition that drives innovation as companies harness AI’s transformative power.

Dig deeper into AI’s Swiss cheese problem with Tim’s takeaways from the conversation.

Do large language models make developers 26% more productive, as a paper recently claimed? Chelsea Troy, leader of Mozilla’s MLOps team and a lecturer in computer science at the University of Chicago, isn’t so sure. There’s so much hype right now—we need voices like Chelsea’s to cut through the noise and help us think a little bit more deeply about what programming actually requires. And that’s just what she did on this episode of Live with Tim O’Reilly.

Chelsea joined Tim for a conversation on why widely cited productivity studies like the “26% paper” are fundamentally flawed, how the real work of programming—reading and analyzing existing code rather than writing from scratch—hasn’t changed, and what skills developers actually need to thrive in an AI-augmented world. Above all, they thought through the challenge of how developers can harness AI’s power without becoming overly reliant on it, which could erode the very cognitive skills that make them effective problem-solvers in the first place.

It’s no surprise that human creativity remains irreplaceable, as Chelsea made clear: Even when assisted by AI, programmers must rely on their investigative abilities to ask the right questions, on their evaluative skills to assess solutions, and on their innovative thinking to push beyond existing patterns. “Machine learning models, the way they are built, the way they are architected, the way they are trained, the way they fundamentally work is to replicate patterns,” she explained. “They’re not designed to generate new patterns, which is exactly what you need in order to move the state of the art.”

Steve Wilson’s day job is at cybersecurity firm Exabeam, where he’s the chief AI and product officer. He also founded and cochairs OWASP’s Gen AI Security Project and wrote The Developer’s Playbook for Large Language Model Security. So he knows a thing or two about the ways AI is creating entirely new attack surfaces that most developers don’t even know exist. On this episode of Live with Tim O’Reilly, Steve shared some eye-opening warnings about the security risks we may be missing.

Steve and Tim started with the hidden dangers of “vibe coding”—the emerging practice of using AI to generate software. “We are in an AI arms race at this point,” he explained, where “every hacker group from Russia to North Korea to a script kitty in his basement in South Africa now has access to world-leading AI.” The conversation then turned to how hackers are exploiting AI hallucinations to convince unwary devs to open malicious packages, the dangers of exposing secrets while livestreaming to thousands of viewers (they’re called “secrets” for a reason), and why running “open source” AI models locally might actually bring foreign adversaries inside your firewall.

But it wasn’t all risks. Steve also discussed practical methods for defending against sophisticated AI-enabled attackers, like using ChatGPT to audit code security and integrating traditional security scanners with AI development workflows. Along the way, he and Tim explored the emergence of new roles like “agent engineers,” and Steve weighed the relative security strengths and weaknesses of various major AI players.