---
source_url: https://doi.org/10.3390/educsci15111537
ingested: 2026-08-03
sha256: 0027950ad90d1183ddb7ac5bdaf81e74f2bbb570f93e327ce7b3a393dbcdff37
---

# Beyond Detection: Redesigning Authentic Assessment in an AI-Mediated World

Steven Kickbusch, Kevin Ashford-Rowe, Andrew Kemp, Jennifer Boreland, Henk Huijser (Learning and Teaching Unit, Queensland University of Technology). *Education Sciences* (MDPI), 15(11), 1537. Published 14 Nov 2025. DOI: 10.3390/educsci15111537. Open access (CC BY 4.0).

## Abstract (summary)

Rapid uptake of generative AI is disrupting conventional notions of authenticity in assessment. The dominant response — surveillance and AI detection — misdiagnoses the problem: in an AI-mediated world authenticity cannot be policed into existence, it must be redesigned. Reconceptualises authenticity as constructed in contexts where AI is expected, declared, and scrutinised; the emphasis shifts from what students know to how they apply knowledge, make judgement, and justify choices with AI in the loop. Offers "design for learning" moves: discipline-agnostic patterns positioning AI as a collaborator rather than a cheating application. Two contributions: (i) a conceptual account of authenticity fit for an AI-mediated world; (ii) actionable discipline-agnostic design patterns.

## From threat to tool (3.1)

- Discourse framed via academic integrity/plagiarism risk casts AI as an external threat to be policed (Cotton et al. 2023); detection-led responses face validity and fairness limits (bias against non-native writers), notable error rates, corrode trust, distract from assessment design (Liang et al. 2023; Weber-Wulff et al. 2023)
- Detection = limited, situational tool, not strategy of first resort; re-centre "design for learning"
- Shift: from "how do we prevent AI use" to "how do we enable thoughtful, responsible, effective use in contexts mirroring future work"; excluding AI from assessment creates an inauthentic scenario
- The authentic professional justifies when/how/why they use tools and critically evaluates outputs

## Embedding tools in authentic tasks (3.2)

- Principle: not "include the tools students might use" but "design tasks where tool use is purposeful and aligned with intended learning outcomes"
- Examples: journalism — critique and edit an AI-generated news brief (identify bias, refine clarity, uphold ethics); engineering/architecture — digital fabrication integrated into design projects, focus on decisions balancing efficiency/feasibility/sustainability; health sciences — evaluate AI-assisted diagnostic recommendations, criteria emphasising clinical judgement, patient-centred reasoning, ethics
- Assessment focus shifts to how students CURATE technology use, not whether they can replicate outputs GenAI already provides

## Critical engagement and digital discernment (3.3)

- Digital discernment: capacity to question assumptions, identify biases, make informed choices about when/how to integrate tool outputs (UNESCO 2023)
- AI-generated market analysis may appear coherent but rest on biased training data, omit context, overstate causal claims

## Core principles for modern authenticity (4.1)

- Authenticity as a multidimensional continuum, not a binary, across four intersecting dimensions:
  1. **Task-context alignment** with contemporary professional practice (not superficial replication; judgement, decision-making, problem-solving under uncertainty)
  2. **Foregrounding professional judgement and ethics** (sustainable assessment, Boud & Soler 2016; UNESCO 2023; collaborative assessment — Boud & Bearman 2024)
  3. **Visibility of process** (iteration, critique, rationale) — polished outputs can mask superficial understanding; process orientation reflects the "messiness" of authentic professional tasks (T. Herrington & Herrington 2006)
  4. **Appropriate use of tools** (including AI) within human decision-making — balance tool use with human creativity; tools as enablers, not substitutes (Kasneci et al. 2023; Mao et al. 2024)
- De-escalate AI anxiety while doubling down on critical thinking (long-division-to-calculator analogy); conceptual understanding and tool competence can develop in tandem
- Stage-appropriate authenticity: early units constrained, well-scaffolded tasks; later units open complexity, uncertainty, stakeholder engagement

## Examples by discipline (4.2)

- Engineering/design: prototype + reflection on design choices, trade-offs, ethical considerations (sustainability, accessibility, safety)
- Education: pre-service teachers evaluate AI-generated lesson plans — focus on capacity to assess inclusivity, curriculum alignment, pedagogical soundness; propose modifications, justify in terms of learning outcomes, consider ethical questions of classroom AI reliance
- Creative industries: integrate AI-generated images/text into human-led workflows; document creative decision-making, rationale for modifications, reflect on human-machine creativity interplay
- Healthcare: appraise AI decision-support patient care suggestions with clinical judgement
- Business/Finance: interrogate a chatbot-generated market analysis; verify claims, identify missing context

## Designing for transparency and reflection (4.3)

- Process logs, AI prompt records, "behind the scenes" artefacts alongside final output (drafts showing iterations, logs of prompts and revisions)
- Short reflective commentaries: explain key decisions, justify tool use, account for changes; integrated into assessment design with explicit criteria for depth, criticality, ethical awareness (Villarroel et al. 2018)
- Oral defences, annotated portfolios, recorded walkthroughs — probe student thinking in real time, reflect professional practices (pitching, consultation, peer review)
- Self-critique and peer feedback strengthen transparency and collaboration (Arthars et al. 2024; Boud & Molloy 2012 — feedback literacy)
- Progressive release across a programme: early units = transparency artefacts + short oral defences; mid = collaboration, negotiated briefs, peer review; capstones = external stakeholders, open-ended briefs, negotiated criteria, explicit ethical framing

## Challenges (5)

- **Equity and access**: not all students have equal access to advanced tech (Selwyn 2021); institutionally provided AI access (fenced deployments) reduces back-channel inequality; equity also about disability, linguistic/cultural diversity; authentic formats can create NEW barriers (workload, employment/carer constraints, real-world simulation limits) — mitigate via transparent workload modelling, staged scaffolding, equivalent-standards modality choice, accessible alternatives (Fawns et al. 2024)
- **Ethics, bias, responsibilities**: tools trained on culturally specific datasets reproduce inequalities/stereotypes; outputs fluent yet unfaithful (Bender et al. 2021); institutional responsibilities: PIA/DPIA for AI tools with data-flow maps and student-facing summaries, vet/approve tools against privacy/bias/accessibility/auditability criteria, standardise prompt/process transparency conventions (prompt logs, version control) without disclosing personal data (NIST 2024; OAIC 2020; QAA 2023; TEQSA 2024)
- **Assessment load and feasibility**: process artefacts and oral defences increase workload; needs workload modelling and feasible scaffolds
- **Staff development**: design-led collaboration over superficial digital tool training

## Conclusion

- Move decisively from detection-focused responses to design; detection has not provided sufficient reassurance and cannot address the broader reshaping of knowledge work
- Curriculum: assess how students apply, adapt, justify knowledge in dynamic contexts rather than recall/reproduction
- Policy: institutional resourcing, equity of access, ethical frameworks anticipating bias/misinformation/data governance
- Educators: confidence in rethinking practices, design-led staff development; learning designers as partners (Kickbusch et al. 2025)
- Further research: discipline-specific implementations, student experiences, impact on learning outcomes and employability
- "Moving from detection to design is not a defensive manoeuvre; it is a constructive agenda for preparing students to engage critically and creatively with the tools that will shape their futures."