More output creates more choices
When teams can generate prototypes, documents, and code more quickly, they have more opportunities to explore. They also have more plausible ideas competing for attention. My view is that choosing what deserves investment becomes more consequential. That is an argument about the work, not a prediction that every PM title or team will grow.
Practice asking better questions
What user problem are we solving? What evidence supports it? What would make us change direction? AI can help draft alternatives and challenge assumptions, but someone must decide which uncertainties matter. Strong product practice makes those questions explicit before a polished output creates premature confidence.
Learn to evaluate AI behavior
For products that contain AI, a single happy-path demo is not enough. PMs need to help define acceptable behavior, difficult cases, fallback experiences, and useful measures. Include latency and cost per successful task alongside quality. Work with engineering and design to establish realistic release criteria.
Delegate tasks, keep accountability
A Microsoft study surveyed 885 PMs and examined telemetry and interviews for subsets to understand AI delegation. Its single-company setting limits generalization, but the focus on judgment and accountability is useful. In your own work, specify which tasks AI can draft and which claims, actions, or decisions require human review.
Build a durable learning habit
Pick one workflow, test an improvement, and compare the total effort and quality with your baseline. Keep what helps. Learn technical concepts well enough to ask clear questions and collaborate. Customer understanding, commercial reasoning, and trust remain skills worth practicing as tools change.
Source / further reading: PM AI-delegation research ↗