Where the Editor Model sits on Bloom's ladder

Almost every teacher has seen Bloom's taxonomy — the ladder of thinking that climbs from remembering facts up to creating something new. It has shaped lesson plans and exam papers for seventy years. I want to do one specific thing with it: show exactly where my Editor Model sits on that ladder, and why AI shoves the weight of learning onto a single rung.

A quick, honest refresher

Bloom's original 1956 framework listed six levels as nouns — knowledge, comprehension, application, analysis, synthesis, evaluation (Bloom, 1956). In 2001 a team led by Anderson and Krathwohl revised it: they turned the levels into verbs and, crucially, reordered the top two so the ladder now runs Remember, Understand, Apply, Analyze, Evaluate, and Create at the summit (Anderson & Krathwohl, 2001; Krathwohl, 2002).

What AI just did to the bottom of the ladder

Here is the uncomfortable part. A capable AI can now perform the bottom four rungs on demand. It remembers more facts than any student, explains them (understand), works examples (apply), and breaks arguments into parts (analyze). It will even draft at the very top — it produces things that look like Create. For decades, climbing those rungs was how learning happened. Now a machine vaults them in a second.

This is the same problem the Editor Model started from: when the machine can produce the work, you can no longer read the finished product as proof that a human climbed anything (Yan & Gašević, 2026).

Evaluate becomes the load-bearing rung

So which rung is left standing as distinctly, unavoidably human? Evaluate. The moment you must judge whether the AI's fluent output is true, relevant, well-reasoned, and fit for your purpose, you are doing the one thing it cannot reliably do for itself. In the AI era, Evaluate stops being the second-from-top luxury and becomes the rung that carries the whole structure.

That is exactly what the Editor Model calls adversarial literacy — the trained habit of interrogating AI output and expecting to find the flaw. Bloom gives it a home on the ladder; the Editor Model gives it a method.

Mapping the DRAFT loop onto Bloom

The Editor Model's five-step loop isn't a rival to Bloom — it's a walk up the top of the ladder, with the machine doing the lower rungs so the human can spend their effort where it counts.

• Delegate — you hand Remember, Understand and Apply to the AI. That's not laziness; it's choosing where to spend human attention.
• Read adversarially — Analyze and Evaluate. You take the draft apart and judge every claim.
• Authenticate — Evaluate again, against real sources. Verification is evaluation with evidence attached.
• Forge — Create, but the human kind: the synthesis, argument, or decision the AI could not reach alone.
• Take ownership — Evaluate turned on yourself (metacognition): can you defend every choice and own the consequences?

Notice how the weight sits on Evaluate. Three of the five steps are evaluation in one form or another. That's the shift Bloom's revisers could not have predicted in 2001, but their ladder still holds it perfectly.

Create still matters — it just changes hands

I don't want to knock Create off its perch. It stays at the top. But we have to be honest that the word now covers two different acts. There is AI-Create (generating plausible text) and human-Create (committing to an original judgement you can defend). The Editor Model protects the second by refusing to accept the first without a fight. Create is still the summit; you just have to climb the last stretch yourself.

What this means for a teacher on Monday

• Stop spending assessment marks on the bottom rungs the AI now owns — pure Remember and Understand tasks no longer tell you who learned.
• Move the marks up to Evaluate and human-Create: judging sources, defending choices, spotting the flaw in a supplied AI answer.
• Grade the climb, not just the summit — the annotated log of what the student caught is often better evidence than the polished final piece.