# Why Instructional Designers Should Learn to Challenge AI **Source**: [[3. Resources/Readwise/Articles/Why Instructional Designers Should Learn to Challenge AI]] > [!question] Lingering Questions > - What is the real time cost per design cycle of "arguing with AI," and does it net positive against rework avoided? The article asserts this without data. > - Can the challenge habit be taught to novices directly, or does it require prior instructional-design judgment to recognize a weak AI answer in the first place — a chicken-and-egg problem the piece doesn't address? > - Does the technique's value depend on the AI genuinely self-critiquing, versus just performing contrition when asked "what's weak here?" > [!tip] Article Ideas > - A one-page "prompt then pressure" rubric template, lifting the article's four artifact-specific question sets (sequencing, objectives, assessments, visuals) into a reusable checklist for any AI-drafted deliverable, not just instructional design. > [!link] Cross Reference > - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — the prompting/pressuring distinction generalizes past instructional design to any AI-assisted knowledge-work review process > - [[3. Resources/Permanent Notes/Goodharts Law Becomes Automatic Under AI Tool Use]] — both claims describe a metric or acceptance criterion silently degrading once AI sits upstream of it > [!quote] Vocabulary > - **Prompting vs. pressuring** — the article's central distinction: prompting gets AI output, pressuring tests it (asking for alternatives, self-critique, counterfactuals) before accepting it into real work > - **Passive evaluator drift** — the cognitive-posture risk where a designer slips from actively authoring to passively approving a plausible-looking AI draft, without a competence loss, just an applied-effort loss ## Bib capsule RK Prasad argues that GenAI's fluency at drafting instructional artifacts (objectives, storyboards, assessments, narration) creates a trap: designers start judging AI output by whether it looks usable rather than whether it's the strongest instructional choice, and proposes a five-habit discipline ("arguing with AI") as the countermeasure. ## Author's thesis Prompting an AI produces output; only deliberately pressuring that output — asking for alternatives, forcing self-critique, defending your own reasoning against it, asking counterfactuals, treating polish with suspicion — protects the instructional judgment that keeps a course from inheriting the AI's quiet oversimplifications. ## Key arguments (≤3) 1. **Fluency is not proof of quality** — AI-generated weaknesses in instructional design tend to be subtle (a little too vague, generic, decorative) rather than obviously wrong, which is exactly what lets them slip past a designer optimizing for speed. 2. **The "passive evaluator" drift is a posture risk, not a competence loss** — designers don't get worse at instructional design; they just stop actively applying it once a plausible draft already exists to react to. 3. **The five-habit discipline has a training effect independent of any single project** — designers who practice it get measurably better (per the article's claim, untested) at spotting vague objectives, weak distractors, and oversimplified content in any draft, AI-written or not. ## Claim atoms ### Pressuring AI beats prompting for output quality From the article's "AI Should Not Just Be Prompted" section — the single most transferable line: a rubric or prompt guide can be built around this one distinction alone. ### Fluency is not proof of instructional quality From "Treat Smoothness With Suspicion" — directly counters the default heuristic ("if it looks good, it's fine") the article opens by naming as the danger. ### Shallow AI review lets its weaknesses become the course's weaknesses From the "architect to passive evaluator" argument — the mechanism-level claim explaining *why* the habit matters, not just that it does: reviewing a plausible draft takes less generative effort than authoring from scratch, so shallow review lets the AI's structure and blind spots pass through unchallenged. ## Permanent candidates - **The prompting/pressuring distinction as a review-rubric anchor** — the strongest, most reusable candidate; promoted below. ## Processing checklist - [ ] Check whether "Nemesis AI Prompts" (referenced in the article's related-posts rail) covers the same ground in more tactical detail - [x] Connect into the relevant MOC — [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] ## Links - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — the prompting/pressuring distinction is a generalizable pattern for any AI-assisted knowledge-work review process, not just instructional design