The clinic had two new kinds of silence.
The first was a familiar one; the early morning quiet before the first sessions begin, when the hallway lights feel too bright and the coffee hasn’t kicked in yet.
The second was newer and lived in the documentation.
Alex, a mid-career BCBA, noticed it while scanning the week’s notes. Everything looked clean. Paragraphs were well-formed. Grammar was flawless. Sentences had the smooth, slightly polished tone of something written by a professional who never had to juggle three cancellations, a late parent call, and a client who woke up sick.
The notes were “good.” They were also…flat.
Same cadence. Same transitions. Same clinical phrasing. Everything started to sound like it came from one voice. But it was supposedly capturing different clients, different teams, different days.
Alex didn’t assume misconduct. Nobody was trying to falsify. Everyone was tired. The tool made the job easier, and leadership had reinforced “done by 6 PM” as the gold standard. Those are all good things, right?
But that’s how Idiocracy dynamics show up in real systems. Not as sabotage, as drift. Convenience gets rewarded long enough that reflection begins to feel optional.
Alex called a quick supervision huddle.
“Show me how you’re using the AI tool for notes,” Alex said.
The first RBT pulled up the interface and shrugged. “I just paste in my bullets and hit generate.”
Another added, “I used the prompt from the training. It’s saved. It works.”
Alex stared at the saved prompt, highlighted in a little template box like it was a finished product:
“Write a clinical session note summary. Make it professional. Use ABA language.”
That wasn’t a prompt.
That was a wish.
And it explained the absence Alex could read in the documentation. The tool was being treated as a vending machine. Put in a token. Get out a note. No iteration. No checking. No shaping.
In a field that literally exists to analyze contingencies, that should feel strange.
Prompting Is Shaping
Here’s the pivot we want you to make.
