# ChatGPT and AI Literacy — How Students Learn Through Use > [!question] Lingering Questions > - Does repair literacy transfer across tools (ChatGPT → coding assistant → search) or stay tool-specific? > - How do these five use-modes differ for multilingual students or students with disabilities — the underlying study doesn't say? > [!tip] Article Ideas > - Map the five-use taxonomy (workhorse, repair, companion, metacognitive partner, trust calibration) onto how contributors actually use AI coding assistants — a direct DevRel/Training crossover. > [!link] Cross Reference > - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] > [!quote] Vocabulary > - *Repair literacy* — the skill of noticing a vague, wrong, or off-task AI answer and rephrasing, adding context, or challenging the tool instead of accepting it. > - *Genre portfolio management* — the meta-skill of fluidly switching AI-use modes depending on the task at hand. ## Bib capsule Educators Technology, "ChatGPT and AI Literacy: How Students Learn Through Use," summarizing Ammari, Chen, Zaman & Garimella (2026), *Learning to live with AI*, arXiv. [[3. Resources/Readwise/Articles/ChatGPT and AI Literacy How Students Learn Through Use|ChatGPT and AI Literacy - How Students Learn Through Use]] ## Author's thesis Students are already developing AI literacy through naturalistic daily use — not primarily through formal lessons — and a study of 10,536 real ChatGPT messages from 36 undergraduates identifies five recurring use modes; the deciding factor in whether a use is educational is what the student does *after* the answer appears, not the answer's quality. ## Key arguments 1. Five uses recur across a year of real student-AI interaction: academic workhorse, repair and negotiation, emotional companion, metacognitive partner, and trust calibration. 2. The locus of learning is post-answer student behavior — checking, applying, repairing — not the AI's output itself. 3. AI literacy is relational as well as technical: students manage trust, frustration, confidence, dependence, and control across these modes, and naming the modes lets them choose deliberately (genre portfolio management). ## Claim atoms ### Learning lives in what students do after the answer > [!quote] ID:rw_h1 > The difference lies in what the student does after the answer appears. The single most reusable claim in the piece: relocates the pedagogical target from AI output to student follow-up behavior. ### Repair literacy inoculates against weak answers > [!quote] ID:rw_h2 > Students who can repair an AI interaction are far less likely to accept weak answers at face value. Names a teachable, assessable skill distinct from prompt-writing. ### Emotional reliance risks dependence on agreeable AI > [!quote] ID:rw_h3 > The risk comes when reassurance turns into dependence, or when students accept overly agreeable responses because the tool makes them feel better. The key risk framing for the "emotional companion" use mode. ### AI literacy is relational, not only technical > [!quote] ID:rw_h5 > AI literacy is technical, but it's also relational. Reframes literacy to include managing trust, frustration, confidence, dependence, and control — not just tool proficiency. ## Permanent candidates - "Learning lives in post-answer behavior" — promoted. - "Naming AI-use modes enables deliberate choice" — promoted. ## Processing checklist - [ ] Read the Ammari et al. arXiv primary source directly and log per-use frequencies before citing specific percentages - [ ] Note the study's limits explicitly when reusing (36 students, one year, apparently one institution) - [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]] — AI-use literacy as a knowledge-work skill, not just an education-sector concern.