How I think about learning work
A few positions that shape how I work. Each one links to a case where you can see it in practice.
Content is getting cheap. The value moves to diagnosis, practice and honest measurement.
With AI, a decent first draft of course text is no longer the hard part. What still takes judgment is understanding what people need to do differently and designing the practice that gets them there. I now spend more of my time on that, and I try to measure whether the learning worked rather than whether people reached the end.
See it in practiceAn AI-assisted design stage for course productionAI adds the most value in design, not in writing the text.
Generating text is the obvious use of AI, and it mostly speeds up a step that was already fast. The bigger gain comes earlier, when AI helps work through source materials, interviews and the brief to find the behavior gap. That’s why I put AI into the design stage, before anything gets written.
See it in practiceAn AI-assisted design stage for course productionA course without safe practice informs; it doesn’t teach.
Reading a rule and applying it in a real situation are different skills. People need a place where they can make the wrong call and see what happens without real damage. Even in mandatory compliance training, I look for room to add situational tasks around the required content.
See it in practiceA role-based compliance course on unfair market practicesTest the riskiest assumption first, and cheaply.
Most learning initiatives rest on one or two assumptions nobody has checked. If such an assumption is wrong, everything built on it goes to waste. In the personalization research I proposed testing the hypotheses on mockups and a small group before building any data infrastructure.
See it in practiceResearching learning personalization before building itA rough first draft moves a project faster than another meeting.
It’s hard for stakeholders to react to an abstract plan and easy to react to something they can see. I’d rather bring a rough prototype early and let people point at what’s wrong. With AI, such drafts are cheap enough to throw away.
See it in practice · TODO
The LMS is turning into an open ecosystem.
Learning no longer has to live inside one closed platform. Standards like xAPI let tools outside the LMS report what people actually did. That changes both what we can measure and what we can build.
See it in practice · TODO