AI Makes Content Cheap. Design Makes It Useful.
The interesting question is no longer whether AI can generate learning content, but whether the generated material helps people make better decisions.
- AI literacy
- Learning design
- Judgment
The first wave of AI in learning was a production story. Faster outlines, faster scripts, faster assessment items, faster translations. That story is now settled: generating material is cheap, fast and mostly adequate.
The second wave is harder and more interesting. It is about judgment.
Cheap production changes the bottleneck
When producing a module took six weeks, scarcity did the filtering for us. Only the important things got built. Now that a draft takes an afternoon, nothing filters — so organizations accumulate material faster than people can use it.
The bottleneck has moved from production to decision-making:
- what deserves to exist at all;
- what should be a job aid instead of a course;
- what should be practice instead of explanation;
- what should simply be fixed in the process rather than trained.
Where AI genuinely helps
Used carefully, AI is strong at the parts of the work that are structural rather than judgmental:
- Turning messy source material into a first structure.
- Generating variations for practice — cases, objections, edge scenarios.
- Acting as a rehearsal partner for conversations.
- Compressing and rewriting for a specific audience and reading level.
The most valuable AI use in learning is not writing the course. It is multiplying the number of times someone can practise and get a response.
Where it quietly fails
AI does not know what matters in a specific organization. It does not know which mistake is expensive, which manager will undermine the change, or which five-minute conversation would remove the need for training entirely. It produces confident averages, and averages are exactly what a specific capability problem does not need.
It also flattens voice. In editorial work and in teaching, voice is part of the credibility. Material that sounds like everything else is easy to ignore.
Practical discipline
A few working rules that hold up:
- Use AI on the middle of the work, not the beginning or the end. The framing and the final judgment stay human.
- Never accept generated content you could not defend in a meeting.
- Prefer generating practice over generating explanation.
- Keep a human decision log — what you kept, what you rejected and why.
Design is the scarce resource now. AI simply makes the absence of design more visible, and faster.