Dr Adrian Waite posted his experience of asking ChatGPT how to run an L&D function with absolutely no discernible impact. The reply he got was cringe-worthy and very funny, so I had to try it myself. Sure enough, ChatGPT handed me a parody of corporate learning that was uncomfortably close to real life: compliance-only training, success defined by completion rates, avoiding real business problems, never leaving the LMS, and designing for approval rather than application.
Then I ran the same prompt in an AI chat tool at work — and the difference was eye-opening. Same question, different framing, very different results. The exercise says as much about us in L&D as it does about the models: the satire only lands because so many of the anti-patterns are recognisable in how learning functions actually operate.
If any of the parody feels familiar, that is the useful part. It is a checklist of what to stop doing: measuring activity instead of impact, shipping training “just in case”, and treating the LMS as the boundary of learning. The full article walks through both outputs side by side.