I just saw an advertisement for a user research tool that mocked people for manually coding research insights when AI agents can do it automatically.

How tragic it is to not understand that often the value lies not in the task getting done but in the process of doing it. In the doing, we learn. We gain context and understanding, and it is from that context and understanding that disruptive insights emerge.

For decades, we’ve struggled with this unspoken, unquestioned belief that more work in the done column is inherently a good thing. I don’t think it is. A high Jira ticket completion rate tells me nothing except that you’ve moved a lot of bits and bytes through Atlassian’s servers.

The Agile manifesto addresses this confusion between output and outcomes directly in its tenth principle: “Simplicity—the art of maximizing the amount of work not done—is essential.” Agile values mindfulness—understanding the present, with its context of what you learned yesterday, and prioritizing what is most valuable given everything you know and have learned. Inevitably, work we scope at the beginning of a project becomes less important, even unimportant, as we work. In the doing, we learn, and what matters changes.

If we fully offload the doing to AI, we get the satisfaction of a pile of completed tasks, but at what cost? AI can check the box on the to do list, but will it have the eureka moment as disparate threads suddenly weave into a rich tapestry? Will it tell you if it does?

Don’t get me wrong. AI is a powerful tool. Agents will be—and already the are being—revolutionary in so many fields and industries.

But for those of us who create value by solving problems, a little skepticism about AI is a good thing. If we leverage AI to do the work we need to do to learns we can end up painfully mediocre.

It’s in the doing, not the done, that we find creativity, innovation, and disruptive ideas.