It feels like every week there’s a new “AI-powered” tool promising to automate our jobs, from writing user stories to synthesizing research. While these tools are great for productivity, I’m noticing a more profound shift in expectations for PMs. It’s no longer enough to just use AI; we’re now expected to have a deeper “AI literacy.”\n\nI’m talking about understanding the fundamentals beyond the slick UI. Things like the difference between various models, the importance of training data, and the inherent limitations and biases of the technology. Why? Because our engineers, stakeholders, and leadership are looking to us to make sound strategic bets on AI features. We can’t do that if we treat the technology like a black box.\n\nBuilding an AI-driven product requires a different kind of product sense. We need to be able to assess technical feasibility, anticipate weird edge cases, and guide our teams on a path that’s genuinely innovative, not just a gimmick. This shift feels as significant as when PMs first had to get comfortable with agile methodologies or basic API concepts.\n\nSo, where do we draw the line? How much technical depth in AI and machine learning is truly necessary for a product manager to be effective today?