AI Isn't the Acronym You Think It Is

AI isn't the acronym you think it is.

AM Bhatt, founder of dae, wrote an opinion piece for District Administration titled, “Are we teaching judgement adequately?” His argument is straightforward and uncomfortable: we've spent decades building systems that remove ambiguity from educators' paths — and in doing so, we've quietly removed the conditions where judgment actually develops.

The AI policy rollout is his sharpest example. Districts responded to a genuinely hard moment with reasonable tools: usage guidelines, integrity frameworks, approved lists. What most didn't do was ask students — or teachers — to wrestle with the hard questions themselves. The guardrails went up before the thinking happened.

The most important leadership move here isn't a new framework. It's knowing when to put the pen down. --AM Bhatt, dae

AI needs two more letters. They both start with A and I.

Everyone is talking about AI. Fewer people are talking about what AI actually requires from the professional using it.

Here's the argument we'd add to Bhatt's: AI doesn't automatically build discernment. It can accelerate it — but only under two conditions that no algorithm provides. Awareness. And intention.

Those are the other AI. And they matter more than the first one.

Awareness is what you bring to the output. Not passive consumption of what the model generated, but active recognition of what's actually in front of you — what it assumes, what it omits, what it got close to right. Evidence-first AI can dramatically accelerate awareness, because the evidence behind the output is right there, inspectable, not locked inside a black box. When a lesson plan surfaces alongside the student work samples and classroom recordings that shaped it, you can see the reasoning. You can agree with it, push back on it, revise it. The AI hasn't replaced your judgment. It's given your judgment something real to work with.

Intention is what no AI can supply. It's the professional deciding what this student needs, in this moment, given everything they know that didn't make it into the prompt. Intention is irreducibly human. It's the difference between a tool that informs a decision and a tool that makes one.

This distinction matters because of what most AI-assisted lesson planning actually looks like today.

Two versions of the same tool.

A teacher opens an AI platform and types: Create a 45-minute lesson on figurative language for 7th graders. The platform generates something. It's coherent. It might even be good. But it was built from nothing — no record of how this class responded last Tuesday, no sense of which students are still miscuing on inference, no connection to the writing samples sitting in the teacher's folder. It's a plausible lesson for a hypothetical classroom.

Now imagine a different version. The AI has access to a recent lesson recording, the exit tickets from the week before, and a set of annotated student writing samples. It surfaces patterns the teacher may have noticed — and a few they hadn't. The suggested lesson isn't a generic template. It's a response to what actually happened, built on evidence the teacher can see and interrogate.

The output might look similar. The process is entirely different. In the first version, the teacher is a prompter. In the second, they're a practitioner — using AI the way a skilled clinician uses diagnostic data. The tool accelerates awareness. The professional supplies intention.

That's not a product pitch. It's a design principle.

The compliance trap Bhatt names — and where it leads.

Bhatt's essay is really about what happens when systems solve for their own discomfort with ambiguity rather than for the educator's capacity to navigate it. Fidelity checks against High-Quality Instructional Materials. Monitoring protocols. Look-fors that tell a teacher exactly what the lesson should contain before they've looked at their students.

These tools exist for real reasons. But in accumulation, they do something Bhatt names precisely: they move responsibility away from the professional and toward whoever wrote the last document. Compliance becomes the standard. Discernment becomes an afterthought.

The same risk lives inside AI, if we let it. An AI that generates a lesson plan without evidence isn't a fidelity check — but it produces the same result. The teacher follows a plausible set of instructions built by something that didn't know their students. The locus of judgment has moved again, just in a different direction.

Evidence-first AI pushes back against that drift. It keeps the evidence visible. It keeps the teacher in the position of the one who decides. It makes awareness available faster — and then it waits for the professional to bring intention to what they're seeing.

That's not a small distinction. It's where the culture either shifts or doesn't.

What Bhatt gets right — and what comes next.

His closing line is worth repeating: "The most important leadership move here isn't a new framework. It's knowing when to put the pen down."

We'd extend that. For AI, the most important design move isn't a smarter model. It's building tools that give educators something real to think with — and then trusting them to think.

Awareness without intention is just pattern recognition. Intention without awareness is working in the dark. Evidence-first AI is what makes both possible in the same professional moment.