UNBLOOMS™ / EXPLORE THE FRAMEWORK
Watch the hierarchy fall away. Keep the cognitive work. Put the problem—and human judgment—at the center.
Explore the preprint’s figures ↓An interactive interpretation of Beyond the Pyramid by Tina Austin, Jason Gulya, and Michelle Kassorla.
A familiar representation of the revised taxonomy—not Bloom’s original drawing.
ENTER ANYWHERE. RETURN AS NEEDED.
QUESTION / METACOGNITIVE CHECKPOINT
Surface assumptions, decide what counts as evidence, and choose where to begin.
Before drafting: which sources can support this claim, and what do I need to understand myself?
Reflection connects every movement. Choosing to constrain AI or work without it is a deliberate option throughout.
01 / ARCHITECT WITH UNBLOOMS™
UnBlooms™ centers the problem and allows learners to enter wherever the task and their readiness require. Question, Generate, Critique, and Refine recur; reflection is the organizing principle, not an endpoint.
Students lock an approved source list before drafting. They name disciplinary constraints and when AI should not be trusted.
The script interrupts fluent acceptance. Every unmatched citation becomes a visible decision point, with the same scaffold for every learner.
The resolution log becomes the Decision Trail: what changed, why it changed, what evidence guided the decision, and whose cognition did the work.
Grade the operation, not the polish: deliberate checking, rejecting, revising, and explaining.
FROM THE PREPRINT
Beyond the Pyramid: Bloom’s Taxonomy meets Generative AI
Tina Austin, Jason Gulya, and Michelle Kassorla
Conceptual / theoretical preprint
Listen to the instructor-style explanation, pause at each reflective checkpoint, and follow the citation examples. One possible pass through the process is shown. Synthesized female voice.
Reflection is the process
Download video ↓Welcome to the UnBlooms reflective checkpoint process, developed by Tina Austin. This model helps you decide how AI should support a learning task, and when to work without it. Notice the center: agency, disciplinary judgment, and reflection. Agency means that you remain responsible for your choices. Disciplinary judgment means using the standards of your field to decide what counts as good evidence. Reflection means noticing what you know, what the AI has supplied, and what needs another look. We will follow one possible cycle; you can enter or return wherever your problem requires.
Checkpoint one is Intent and Entry. Before opening an AI tool, ask: what is the learning goal, and should AI enter this task? If you are learning to evaluate sources, for example, your own verification must remain part of the work. Recognize what comes from AI and what you know independently. Identify where support might help. The red Resist branch means you may decide not to use AI at all. That is a reasoned choice.
Checkpoint two is Generate and Verify. You may produce an output or inspect one you already have. Ask what is being claimed, what needs verification, and what is missing. In the Citation Lab, compare draft citations with your approved list. A list match is only a comparison result; it does not prove that a source supports a claim. Check the source itself. Here, identifying a problem leads into analyzing it. You may discontinue AI or seek another approach.
Checkpoint three is Critique and Judge. Look beneath the polished language. What assumptions shape the output, and how accurate is the reasoning? Examine omissions, bias, and the standards being used. In a citation task, ask whether the evidence actually supports the claim, and whose perspective may be missing. Decide what to accept, challenge, revise, or reject, and explain why. The emphasis is on analysis and evaluation. The Resist branch lets you reject this output.
Checkpoint four is Refine and Decide. Use the critique to change the work, not simply to polish the wording. Ask: can I do this better? What have I learned? Should AI remain involved? For example, replace an unsupported citation and rethink the claim it was meant to support. Record the evidence that changed your decision. Evaluation can lead to a better method or solution. You may also remove AI and continue independently.
Checkpoint five is Create or Resist. Apply what you have learned by building a better solution, improving a workflow, or choosing a human approach. The goal is not simply to produce more content. It is to make a defensible decision that reflects your understanding. Explain what you accepted, what you changed, and what you chose not to delegate. Choosing to work without AI can be as deliberate as choosing to use it.
The arrows now return us to the problem. The intended outcome is more deliberate judgment and improved capability, not a guarantee that one pass proves learning. The engagement levels help you notice recognizing, identifying, analyzing, evaluating, and creating or resisting. They are lenses within the cycle, not steps everyone must complete in a fixed order. Use the chapter buttons to revisit a checkpoint, pause to answer its reflective question, and keep a decision trail. Ask yourself: what did I verify, what changed my thinking, and why? That is how the process makes human judgment visible.
Pause at each checkpoint to consider agency, disciplinary judgment, and whether to use AI, constrain it, or continue independently.
Figure 5 · Tina Austin (2025a) · CC BY 4.0 · Animated adaptation of the supplied preprint figure.
The pyramid transformation and reflective process video are new visual interpretations of the framework. The original figures are available unchanged. Figure captions and attribution follow the supplied preprint.
WHY I CALLED IT UNBLOOMS™
Bloom’s original taxonomy was a classification of educational objectives—not the colorful, rigid ladder that later came to dominate classrooms. Its evaluative instinct was gradually buried beneath a simplified hierarchy.
Generative AI made that hierarchy impossible to ignore. A learner can now jump directly to a polished “Create” output without doing the cognitive work the product once appeared to represent. If AI provides the elevator to the top of the pyramid, reaching the top can no longer be enough evidence of learning.
UnBlooms™ “unblooms” the pyramid. It replaces a fixed sequence with a recursive, nonlinear, context-dependent loop. Learners may enter at any point; what matters is whether they question, generate, critique, refine, and make deliberate choices about when AI supports learning—and when it should be resisted.
UnBlooms™ brings judgment back to the center—alongside human agency and curiosity—where the work of learning can be observed, discussed, and assessed.
Read the full origin story on Substack ↗READ THE RESEARCH
Open-access research on agentic assignments, metacognitive infrastructure, and the UnBlooms™ measurement framework.
Tina R. Austin
Journal of Instructional Design and Technology, 1, 8–18 (2026)
Tina R. Austin
International Journal of Emerging and Disruptive Innovation in Education: VISIONARIUM, 4(1), Article 7 (2026)
Tina R. Austin, Jason Gulya, and NICK Potkalitsky
International Journal of Emerging and Disruptive Innovation in Education: VISIONARIUM, 4(1), Article 6 (2026)
THE UNBLOOMS™ ASSESSMENT SHIFT
GenAI weakens the link between a polished product and evidence of learning. UnBlooms™ makes the learner’s decisions, revisions, and resistances assessable.
The UnBlooms™ Workbook provides practical tools for assessing human reasoning in the age of AI.