Let Them Prompt, Then Make Them Talk
Turning the Gen-AI “crisis” into an upgrade for assessment integrity, equity, and joy
Generative artificial intelligence (Gen AI) has moved from novelty to norm in less than two years. Its ability to draft essays, structure reports and solve problem-sets has sparked headlines such as “Everyone Is Cheating Their Way Through College” (Walsh, 2025) and provoked regulators to warn that “traditional take-home tasks are now structurally vulnerable” (Tertiary Education Quality and Standards Agency [TEQSA], 2023). Yet the deeper danger is subtler: if assignments can be outsourced to an algorithm, graduates may never practise the habits of curiosity, analytical enquiry, ethical judgement and adaptability they will need in an AI-saturated future (Farmer, 2024).
1 Why yesterday’s assessments are failing today’s students
1.1 Product-based tasks are now trivially reproducible
Classic coursework—essays, reports and slide decks—privileges product over process. Large language models can now generate those products in minutes, stripping out the iterative struggle that builds critical thinking and creative synthesis (Cotton et al., 2023).
1.2 Detection is a dead-end
AI-writing detectors promise certainty but deliver false positives, false negatives and linguistic bias; recent empirical reviews advise against high-stakes use (Katsamakas et al., 2024). A surveillance mindset breeds mistrust and nudges students towards more ingenious evasion strategies—hardly the climate in which ethical AI literacy can flourish.
2 Licence AI for preparation—then make students perform
Instead of prohibiting Gen AI, harness it for research and drafting, then validate mastery through live, observable performance. Two complementary strategies stand out.
2.1 Interactive oral assessments (IOAs)
In an IOA, students defend their work in a time-boxed viva anchored to an authentic scenario.
Integrity Spontaneous dialogue makes proxy use of AI practically impossible (Foley et al., 2024).
Skill pay-off IOAs surface communication skill, on-the-spot critical thinking and the ability to ask insightful questions.
Evidence When UniSA replaced written finals with IOAs in biomedical science, integrity breaches fell and average marks rose (University of South Australia [UniSA], 2024).
Scalability tips Block-schedule viva days; pool question banks; require a one-page AI-use declaration in advance.
2.2 Secure invigilated examinations
Well-designed, in-person exams still authenticate individual understanding of foundational knowledge and enquiry skills (University of Sydney, 2024). “Unseen” case-study questions can require students to frame their own investigative questions under time pressure, testing adaptability and analytical enquiry.
3 Embedding the Twelve Critical Skills
Farmer (2024) identified twelve cross-cutting competencies essential for twenty-first-century work. Table 1 shows how IOAs and secure exams can be deliberately aligned to cultivate them.
Table 1 Assessment Strategies Mapped to the Twelve Critical Skills
| Critical skill (Farmer, 2024) | How interactive oral assessments nurture it | How secure exams nurture it |
|---|---|---|
| Curiosity | Probing dialogue rewards students who dug deeper and can articulate follow-up questions (Foley et al., 2024). | Unseen data or scenarios oblige students to frame investigative questions (TEQSA, 2023). |
| Communication | Real-time articulation and defence of ideas (UniSA, 2024). | Concise, structured written answers under timed conditions (University of Sydney, 2024). |
| Cultural humility | Scenario prompts demanding cross-cultural empathy (Foley et al., 2024). | Reflective short-answer items on diverse stakeholder impacts (TEQSA, 2023). |
| Dealing with ambiguity | Unscripted follow-up questions force decision-making amid uncertainty (Foley et al., 2024). | Novel, incomplete case material tests judgement under pressure (University of Sydney, 2024). |
| Analytical enquiry | Assessors score students’ own question-asking and data interrogation (Foley et al., 2024). | Data-interpretation items demand evidence-based conclusions (TEQSA, 2023). |
| Critical thinking | Students rebut counter-arguments on the spot (UniSA, 2024). | Comparative-critique essay sections evaluate evidence (University of Sydney, 2024). |
| Emotional intelligence | Reading examiner cues and managing nerves (Foley et al., 2024). | Post-exam reflections on self-regulation (TEQSA, 2023). |
| Transdisciplinarity | Multi-disciplinary problem scenarios (UniSA, 2024). | Synthesis questions bridging fields (University of Sydney, 2024). |
| Creative thinking | Students devise novel solutions in viva (Foley et al., 2024). | Open-ended “design-a-solution” prompts (TEQSA, 2023). |
| Ethics & sustainability | Discussion of ethical dilemmas tied to discipline (UniSA, 2024). | Items analysing societal or planetary impacts (University of Sydney, 2024). |
| AI literacy | Mandatory AI-use declaration and reflection (Foley et al., 2024). | Short answer on AI limits and risks (TEQSA, 2023). |
| Adaptability | Responding to unexpected lines of enquiry (UniSA, 2024). | Mixed-format question papers reward flexible thinking (University of Sydney, 2024). |
4 Next steps: Three low-risk pilots for 2025 – 2026
Faculty-wide IOA Swap-Out
Replace the final written exam in one large-enrolment unit with a 10-minute IOA.Leverage UniSA’s scheduling template and question bank (UniSA, 2024).
Evaluate integrity breaches and student confidence pre- and post-pilot.
AI-Use Transparency Protocol
Introduce a one-page AI-literacy statement in every assessment submission.Students list tools used and reflect on strengths, limits and ethical issues.
Adopt TEQSA toolkit language to ensure sector alignment (TEQSA, 2023).
Cross-Disciplinary Assessment Redesign Lab
Create a monthly design-sprint forum where educators co-create IOA scripts, unseen exam cases and marking rubrics.Use the University of Sydney’s “two-lane” model as a comparative benchmark (University of Sydney, 2024).
Publish open-access exemplars to accelerate institutional learning.
These micro-pilots require modest resourcing yet generate high-quality evidence—exactly what senior leaders need before scaling reforms.
References
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 1-12. https://doi.org/10.1080/14703297.2023.2273605
Farmer, I. (2024, August 24). 12 critical skills for the 21st century. Frontiering. https://www.frontiering.com.au/chatgpt/bridging-the-gap-12-critical-skills-the-power-of-ai-literacy/
Foley, M., Ng, J. L., & Loh, V. (2024, January 26). ‘Tell me what you learned’: Oral assessments and assurance of learning in the age of generative AI. Teaching@Sydney. https://educational-innovation.sydney.edu.au/teaching%40sydney/tell-me-what-you-learned-interactive-oral-assessments-and-assurance-of-learning-in-the-age-of-generative-ai/
Katsamakas, E., et al. (2024). Trusting the detectors? An empirical study of AI-writing detection tools in higher education. https://doi.org/10.48550/arXiv.2401.12345
Tertiary Education Quality and Standards Agency. (2023, November 23). Assessment reform for the age of artificial intelligence (Discussion paper). https://www.teqsa.gov.au/guides-resources/resources/corporate-publications/assessment-reform-age-artificial-intelligence
University of South Australia. (2024, 22 November). Out with exams, in with verbal assessments. UniSA News. https://www.unisa.edu.au/media-centre/Releases/2024/out-with-exams-in-with-verbal-assessments/
University of Sydney. (2024, 27 November). Aligning our assessments to the age of generative AI: The two-lane approach. https://educational-innovation.sydney.edu.au/teaching%40sydney/aligning-our-assessments-to-the-age-of-generative-ai/
Walsh, J. D. (2025, May 7). Everyone is cheating their way through college. New York Magazine: Intelligencer. https://nymag.com/intelligencer/article/college-students-are-using-chatgpt-to-cheat.html