{"resourceId":"the-campus-manchester-ai-authentic-engineering-do-augment-challenge-defend-20261030","versions":[{"version":"external-90e3ec5c3d61ad0059f84299bd230aa9686483b555408608a1da7cde769d03be","resource":{"id":"the-campus-manchester-ai-authentic-engineering-do-augment-challenge-defend-20261030","title":"Manchester engineering academic replaces AI-proofing with a do, augment, challenge, defend sequence, citing ChatGPT accuracy falling to about 41% on open-ended engineering problems","organization":"The University of Manchester (chemical engineering teaching)","sector":"Higher education engineering teaching and assessment design","geography":"Manchester, United Kingdom","publishedAt":"THE Campus, 30 October 2026","publicationDate":"2026-10-30","eventDate":null,"sourceName":"Times Higher Education Campus","sourceLabel":"Practitioner commentary","sourceUrl":"https://www.timeshighereducation.com/campus/difference-between-aiproof-and-aiauthentic-engineering-education","evidenceClass":"independent-reporting","outcomeClass":"emerging","topics":["knowledge-work","operating-model"],"finding":"Salman Shahid (University of Manchester) writes that his research found about 95% of ChatGPT's answers to knowledge-recall chemical engineering questions were correct, falling to about 41% on complex, open-ended problems needing professional judgement, with some wrong answers plausible enough that spotting them required real engineering knowledge. He has moved from AI-proofing to AI-authentic learning in four stages: students attempt the problem first, then use AI as a tutor told not to give the answer, then assess the AI output (sometimes a deliberately flawed solution), then defend their decisions. He argues AI literacy cannot be separated from disciplinary knowledge.","sledRelevance":"Offers a concrete assignment sequence for STEM courses that keeps student thinking ahead of AI use, with a cited accuracy gap that explains why verification skills matter.","evidence":"Author's own teaching practice and a summary of his prior research; no evaluation of the four-stage approach.","architectureImplications":"Interpretation: none specific; the approach works with general-purpose AI tools given a constrained tutor prompt.","governanceImplications":"Implies course rules that permit AI only after an initial student attempt and require critique of AI output.","securityPrivacyImplications":"Not addressed in the source.","caveats":"Single-author practitioner piece; accuracy figures come from earlier research not detailed here; no outcomes; published in a section created in partnership with a sponsor.","streamIds":["student-success"],"roles":{"sales":"Sales framing: this practitioner piece gives engineering deans a practical alternative to the AI-proofing arms race: students attempt first, use AI as a restricted tutor, critique its output and defend their choices. The cited accuracy drop from about 95% to about 41% on open-ended problems makes the case for verification skills. A bounded engagement could adapt the sequence for a few courses. Do not claim it improves learning; the author reports no evaluation.","engineering":"Engineering framing: the approach needs no special platform, but a reusable tutor prompt that refuses to give final answers, and a bank of deliberately flawed AI solutions, are worth building. Check whether campus AI tools allow instructor-set system prompts or shared assistants. Proof of value should test whether the restricted tutor prompt actually withholds answers under student pressure, and whether flawed-solution tasks discriminate between students who understand the principles and those who do not.","delivery":"Delivery framing: delivery means turning one academic's practice into a supported course pattern. Work includes templates for each of the four stages, a shared tutor prompt, a bank of flawed AI solutions per discipline, and a rubric for the defend stage. Teaching centres own the kit; departments own adoption. Evaluate it properly, comparing student performance on open-ended problems with prior cohorts, before recommending it widely."},"roleDetails":{"sales":{"customerProblem":"AI-proofing assessment is a losing battle, and students accept plausible but wrong AI answers on open-ended problems.","stakeholders":["Engineering dean","Department heads","Teaching and learning center","Accreditation leads"],"discoveryQuestions":["How do engineering courses currently handle AI use?","Could students spot a flawed AI solution in your discipline?","Do instructors have time to build critique tasks?"],"valueHypothesis":"Interpretation: an attempt-first, critique-and-defend sequence may build judgement students need to use AI safely in professional work.","potentialEngagement":"Adaptation and evaluation of the four-stage sequence in a small set of engineering courses.","unsupportedClaims":["Learning gains from the sequence","That the cited accuracy figures hold for current models","Applicability outside engineering"]},"engineering":{"fit":"Courses using general-purpose AI tools with instructor-configurable assistants.","architectureIntegration":"Shared restricted-tutor assistant configured on the campus AI platform plus a repository of flawed solutions.","prerequisites":["Campus AI tool with shared assistants","Discipline experts to build flawed solutions","Defence rubric"],"constraints":["Model behaviour changes over time","Students can use unrestricted tools","UK practitioner context"],"security":"Keep flawed-solution banks and rubrics in access-controlled course spaces.","proofOfValue":["Restricted prompt withholds answers under pushback","Flawed-solution tasks discriminate understanding","Compare open-ended problem scores with prior cohorts"]},"delivery":{"work":"Package the four-stage sequence, build a shared tutor prompt and flawed-solution bank, and run an evaluated pilot.","dependencies":["Instructor time","Campus AI platform configuration","Assessment office input"],"ownership":"Teaching and learning center with engineering departments.","skillsAdoption":"Instructor workshop on building critique tasks; student briefing on using AI as a restricted tutor.","governanceCheckpoints":["Assessment rules updated for AI-after-attempt","Mid-pilot review","Evaluation before scale-up"],"acceptanceCriteria":["Proposed: pilot in an agreed number of courses","Proposed: measurable comparison on open-ended problems","Proposed: student feedback on fairness collected"],"risks":["Students skip the attempt stage","Instructor workload","Outdated accuracy assumptions"]}},"retrievedAt":"2026-10-31T03:19:28Z","enrichedAt":"2026-10-31T03:24:06Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Instructors need disciplinary expertise and time to prepare flawed AI solutions and run defence activities.","procurementImplications":"None.","operatingModelImplications":"Interpretation: teaching centres can package the four-stage sequence as a reusable assignment pattern.","sourceVerification":{"openedUrl":"https://www.timeshighereducation.com/campus/difference-between-aiproof-and-aiauthentic-engineering-education","referenceExcerpt":"about 95 per cent of its responses to questions requiring knowledge recall were correct but accuracy fell to about 41 per cent","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}