The Line Moved
Course overview: generative AI collapsed the lower rungs of Bloom's taxonomy, so the line between busywork and learning moved up. What that means for the courses you design, and the one question every lesson returns to.
Who This Course Is For
This course is for the people who decide what a course is — instructors writing the next syllabus, and course designers shaping objectives, activities, and assessments across a program. You don’t need a learning-science background. You need a course, a unit, or even a single assignment you’re willing to rebuild.
Everything here is meta: it’s a course about designing courses. Every module hands you a lens, and you run one of your own courses through it. By Module 5 you’ll have redesigned a real unit you teach.
Start With the Talk
In her TEDxUniversity of Tartu talk, Estonia’s Minister of Education Kristina Kallas opens a lecture about AI by saying she isn’t going to talk about AI — she’s going to talk about humans. Watch it before you go further; it’s the argument this entire course is built to operationalize.
Her thesis, in her own words:
“We have to consciously train humans and as many people as possible to be constantly in the flow of this higher cognitive thinking processes.”
Notice the verb: train. Not “let AI do it,” not “ban AI.” Consciously design for people to spend more of their time thinking hard — which means someone has to design the courses that make that happen. That someone is you.
She frames this as a genuine evolutionary leap, the way the printing press was one: before it, reading was a rare skill; after it, an entire species learned to read. AI, she argues, puts humans under a comparable pressure — but a cognitive one. “We have to think faster, more systematically, and much more creatively.” The pressure isn’t to run faster; it’s to spend far more of our waking hours in higher-order thought than schooling has ever asked of us.
The Line That Moved
Look at Bloom’s Taxonomy of the cognitive domain. It’s a ladder of verbs, from remember at the bottom to create at the top, split by a line into lower-order and higher-order thinking skills.
| Level | Sample verbs | |
|---|---|---|
| Higher order | Creating | generate, design, produce, invent |
| Evaluating | argue, judge, critique, rank, validate | |
| Analyzing | distinguish, contrast, dissect, examine | |
| — the line — | ||
| Lower order | Applying | execute, calculate, demonstrate, translate |
| Understanding | summarize, explain, classify, describe | |
| Remembering | define, recall, list, name, recite |
For most of the history of formal education, a huge share of graded work lived below the line — recall a definition, summarize a chapter, apply a formula, translate a passage. That work was valuable partly because it was hard for a student to fake. Producing a competent summary was evidence a human had read and understood.
Generative AI erased that evidence. A model will now remember, understand, and apply on demand, instantly, for free. The lower rungs didn’t disappear — they stopped being where the learning is demonstrated. The line between “busywork a machine can do” and “thinking only a human is doing” moved up the ladder.
That’s the evolutionary challenge in the talk’s title. It isn’t that AI knows everything. It’s that the tasks we used as proxies for learning are now the tasks AI does best.
The Design Requirement
Here is the sentence this whole course exists to serve:
A well-designed course in the age of AI makes students think with AI, not let AI think instead of them.
Those are two genuinely different experiences that can look identical from the outside — same tool, same screen, same finished artifact turned in. The difference is entirely in whether the student stayed cognitively responsible for the judgment. Your job as a designer is to build courses where the “let AI think instead” path doesn’t earn the grade.
That is not the same as banning AI, and it is not the same as detecting AI. Both of those are losing battles fought below the line. This course fights above it: design the work so that outsourcing the thinking produces obviously worse work.
The reflex is to read “higher-order thinking” as “keep AI out of the classroom.” That’s the wrong frame, and Kallas rejects it. Estonia’s response wasn’t a ban — it was AI Leap, a national program putting AI tutors into schools, deliberately designed to provoke productive struggle rather than hand over answers. Her reasoning is blunt: without AI, students “leap back to the usual lower cognitive thinking” — the tool is what pushes them up the ladder. As Estonia’s president framed the launch, “it’s not about using AI the most… it’s about how we use it and why.” The goal is students who are more cognitively engaged because the tool is there, not students protected from it. If your redesign just walls off the tool, you’ve dodged the design problem instead of solving it.
Course Objectives
By the end of this course, you will be able to:
- Read Bloom’s taxonomy through an AI lens — identify which of your existing objectives now sit below the line and are cheaply completed by a model.
- Rewrite learning objectives so they target analyzing, evaluating, and creating, without pretending foundational knowledge is optional.
- Diagnose why an assessment breaks under AI, and redesign it so the graded evidence lives above the line.
- Apply the five-step loop (attempt → question → critique → transfer → reflect) to convert an AI-vulnerable assignment into an AI-resilient one.
- Design learning activities for productive struggle, using the principles behind Estonia’s tutor design instead of an off-the-shelf chatbot.
- Redesign one real unit you teach end to end — objectives, activities, and assessment aligned above the line.
The Question We Keep Returning To
Every lesson circles back to one question you’ll learn to ask of any assignment:
If a student ran this straight through an AI, would the result be good — and would they have learned anything?
If the answer is “yes, good, and no learning,” the task lives below the line and needs redesigning. If the answer is “the AI can help but the student still has to do the thinking that earns the grade,” you’re above it. Hold that question. It’s the whole course in one sentence.
Before Module 2, choose one real artifact you own — a single assignment, a unit, or a module of a course you teach. Write down:
- What it currently asks students to do (the actual task prompt or objective).
- Which Bloom’s level it targets — be honest, most existing assignments cluster at understand/apply.
- The AI test: paste the prompt into a model right now. How good is the output, and what did the student learn by getting it?
Keep this. Every module gives you a new lens for the same artifact, and Module 5’s capstone rebuilds it.
An instructor, worried about AI, replaces the take-home essay with an in-class handwritten summary of the assigned reading. Through the lens of this lesson, what's the problem?
Reveal Answer & Explanation
They moved the assessment offline but not up the ladder — a summary is still a lower-order task, so they’ve made the work harder to fake without making it higher-order.
Blocking the tool addresses cheating but not the design requirement. A handwritten summary still targets “understand,” a rung AI does trivially, which is exactly why it felt fakeable in the first place. The fix isn’t to police the summary; it’s to change what’s being asked — for instance, having students critique two competing interpretations of the reading and defend a choice (evaluating), work they’d have to think through with or without a tool beside them.
Up Next
Module 2 reads Bloom’s taxonomy level by level for the AI era — what the machine now does cheaply, what stays stubbornly human, and why the answer is not to skip the foundations.