The Question AI Still Isn’t Asking

Most conversations about AI start with productivity. Write faster, summarize the meeting, generate the image, automate the workflow. Those uses are real. They're also the ones I find least interesting.

The biggest opportunities in my career almost never came from doing familiar work faster. They came from asking a question nobody else thought to ask. At Comcast, we already had years of viewing data. What we didn't have was anyone asking whether it pointed to an entirely new creator economy. That question helped produce one of the first YouTube multi-channel networks. The same data held another signal: real American demand for K-pop, years before it crossed over. We launched a channel for it. The information wasn't hidden either time. Only the question was.

Building Audience Genomics taught me the same lesson at product scale. Organizations sit on huge amounts of customer data, creative work, operational history, and performance records. Most of it gets seen through a narrow set of dashboards and recurring reports. We ask what happened. We ask whether the number went up or down. We ask which campaign won. Those questions are necessary, and they're bounded by what the company already knows to look for.

AI changes that, and not only because it makes existing work faster. It makes it practical to investigate questions that used to be too expensive, too complex, or simply impossible. What traits show up again and again in the content that gets an emotional response? What unrelated-looking customer behaviors add up to an emerging need? What value is trapped in assets a company files under "operational" instead of "strategic"? What assumptions about the business were never tested because no one had a way to test them?

Here's why that distinction matters. When every company has the same models, automation alone won't hold an advantage for long. Everyone will use the same tools to wring out the same efficiencies. The edge will belong to the organizations that get better at knowing what's worth investigating, and that takes more than technology. It takes curiosity, judgment, real domain knowledge, and the willingness to look past the questions the company has always asked. The companies that get the most from AI won't just automate yesterday's work. They'll use it to find tomorrow's opportunities.