Signal Intelligence Briefing 001
Generative AI is weakening the assessment assumptions that many tertiary systems still rely on: that writing shows thinking, that submission shows authorship, and that an assessment artefact reliably evidences capability. The strategic issue is not simply cheating or tool use. It is whether institutions can still verify capability, protect credential trust, and make defensible judgements about learning in an AI-saturated environment.
1. Purpose of This Briefing
Primary Goal
This briefing explores the widening gap between:
current assessment assumptions
operational reality in an AI-saturated learning environment.
Its purpose is not to prescribe immediate solutions, but to surface emerging tensions that may increasingly affect:
assessment validity
moderation confidence
credential trust
workforce signalling
institutional capability
Core Strategic Premise
Most tertiary systems still operate as though:
writing = thinking
submission = authorship
assessment artefact = capability evidence
Those assumptions are destabilising rapidly as generative AI tools become embedded in everyday study practice and professional work. Empirical studies show that markers often cannot reliably distinguish GenAI-assisted work from unaided work, even in “authentic” assessment tasks, and that current assessment approaches remain susceptible to GenAI manipulation (Kofinas et al., 2025; Lodge et al., 2023; Lodge et al., 2025).
The emerging issue is not simply academic integrity, but widening trust ambiguity around what educational evidence actually confirms.
What This Briefing Is Not
This briefing does not argue for:
banning AI
eliminating online learning
universal oral assessment
abandoning existing assessment systems
Nor does it assume all AI use is inappropriate.
Its focus is narrower and more operational:
how institutions maintain trustworthy capability signals under conditions where traditional evidence assumptions are weakening.
Desired Reader Outcome
By the end of this briefing, decision-makers should understand:
the problem is already operational, not theoretical: regulators, universities and peak bodies across Australasia now describe GenAI as a structural driver of assessment reform rather than an isolated integrity issue (Lodge et al., 2023; Lodge et al., 2025)
current responses are fragmented and inconsistent
detection-centric approaches are unlikely to stabilise the system alone
capability verification approaches may become increasingly important
implementation guidance is currently lagging behind institutional reality
2. Executive Signal Summary
Signal 1 — Invisible AI Use Is Already Normalised
In degree-level and professional contexts where learners have regular independent digital access, AI assistance is increasingly embedded in everyday workflows.
In many contexts, usage is:
undisclosed
difficult to detect consistently
unevenly governed
pedagogically inconsistent
The operational reality has moved faster than institutional response systems.
Sector guidance from TEQSA and major Australian universities now explicitly assumes widespread GenAI use by students and staff, and emphasises that institutions must redesign assessment and integrity regimes on that basis (Lodge et al., 2023; Lodge et al., 2025; The University of Sydney, 2024).
Signal 2 — Traditional Assessment Assumptions Are Weakening
Confidence in unaided authorship is eroding, particularly where:
take-home assessments dominate
asynchronous online learning is common
large-scale marking occurs
authenticity checks are limited
This creates increasing uncertainty around what credentials actually verify.
Experimental work in UK and European higher education contexts shows that experienced markers often cannot reliably identify GenAI-assisted submissions, and that authentic assessments alone do not safeguard integrity in GenAI-rich environments (Kofinas et al., 2025).
Signal 3 — Detection-Based Responses Are Unlikely To Stabilise The System
AI detection technologies currently face:
reliability concerns
false positives
rapid model evolution
adversarial behaviour
inconsistent institutional implementation
Detection may remain useful as a supporting mechanism, but appears unlikely to function as a primary trust architecture.
Multiple reviews and sector guidance warn that over-reliance on detection tools is unstable, given false positives, rapid model evolution and adversarial use, and instead recommend structural assessment reform over tool-centric responses (Tertiary Education Quality and Standards Agency, 2023, 2025; Lodge et al., 2025).
Signal 4 — Practitioner Adaptation Is Occurring Unevenly
Some educators are already redesigning:
assessment structures
oral verification processes
staged evidence collection
applied demonstrations
reflective explanation tasks
Others remain:
uncertain
unsupported
institutionally constrained
Adaptation across the sector remains fragmented.
Australian case material highlights substantial local experimentation in assessment design and policy, but also wide variation in readiness, staff capability and coherence across programmes (Lodge et al., 2025; Bridgeman & Liu, 2025; Tertiary Education Quality and Standards Agency, 2025).
Signal 5 — Capability Verification May Become Increasingly Important
A growing number of providers are exploring:
oral verification
live demonstration
conversational confirmation
applied evidence
process visibility
iterative evidence gathering
Capability verification refers to approaches that strengthen confidence that meaningful capability is genuinely present, rather than inferred from a single artefact alone.
Many of these practices already exist within parts of vocational and professional education. What may be changing is their strategic importance under AI-assisted conditions.
Assessment reform guidance now emphasises programme-level assurance of learning, performative and social assessments, and multiple evidence points, rather than relying on single asynchronous written artefacts (Kofinas et al., 2025; Lodge et al., 2025).
3. Why This Matters
A. Credential Trust
If systems cannot reliably distinguish:
assisted production from
demonstrated capability
then confidence in credentials may weaken over time.
International commentary on GenAI in higher education warns that unresolved questions about assessment integrity and verification will progressively erode trust in formal credentials, particularly where programmes are highly online and text-dominant (Kofinas et al., 2025; Lodge et al., 2025).
This is particularly significant in workforce-facing contexts.
B. Workforce Readiness
The issue is not AI use itself.
The issue is whether systems can still reliably determine:
understanding
judgement
transfer capability
communication ability
practical application
Employers ultimately care about capability reliability.
As AI increases the ability to simulate competent outputs, the ability to distinguish between assisted production and reliable applied capability may become increasingly economically important.
Regulatory commentary stresses that assessment must still robustly confirm learning outcomes and workforce-relevant capabilities, even as AI becomes a normal tool in professional practice (Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2025).
C. Equity and Literacy Complexity
AI tools may simultaneously:
support learners and
mask underlying capability gaps
This creates a more complex literacy environment where:
surface performance may not equal independent capability.
Emerging practice notes that GenAI can simultaneously scaffold participation and obscure underlying literacy and communication gaps, complicating institutional responsibilities for equity and assurance of learning (Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2025).
D. Institutional Risk
Without coherent adaptation, providers may face increasing pressure around:
moderation confidence
reputational trust
assessment validity
inconsistent learner expectations
compliance ambiguity
Regulators are signalling more explicit expectations that providers can demonstrate how their assessment regimes manage GenAI-related risks while maintaining standards, increasing both compliance visibility and institutional exposure (Bridgeman & Liu, 2025; Tertiary Education Quality and Standards Agency, 2025).
4. Current System Reality
Observation 1
Many educators privately acknowledge:
AI usage is widespread
existing policies are difficult to enforce consistently
assessment redesign is lagging behind operational reality
Yet institutional responses remain uneven.
Observation 2
Many institutions are currently oscillating between:
Response Type
Primary Weakness
prohibition
difficult to enforce
unrestricted use
weak verification clarity
detection focus
low long-term stability
redesign experimentation
uneven implementation capability
Observation 3
There is currently no widely adopted national operational model for:
AI-era assessment legitimacy
capability verification
authenticity assurance
evidence confidence
This creates increasing fragmentation across the sector.
Across Australasia, guidance on GenAI tends to take the form of high-level principles and voluntary pathways rather than a single operational model, leaving significant room for local interpretation and fragmentation (Australasian Academic Integrity Network, 2023; Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2023).
Regulators such as TEQSA are moving toward more explicit, regulatory-led expectations for how providers manage GenAI-related assessment risks from 2026 onwards, which will sharpen external scrutiny of institutional responses (Bridgeman & Liu, 2025; Tertiary Education Quality and Standards Agency, 2025).
5. What The System May Be Misunderstanding
Misunderstanding 1 — “The Issue Is Cheating”
The deeper issue is:
the weakening relationship between artefact production and independent capability.
This is not simply a misconduct problem.
It is increasingly becoming a systems problem.
Misunderstanding 2 — “More Policy Will Stabilise The System”
Policy alone is unlikely to resolve:
operational verification challenges
assessment design limitations
capability interpretation problems
Practical implementation models may matter more than policy expansion alone.
Recent regulatory papers emphasise that policy settings alone are insufficient; institutions must develop concrete assessment plans, curricular redesign, and assurance mechanisms at programme and course level (Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2023, 2025).
Misunderstanding 3 — “AI Use Removes Learning”
Reality is more nuanced.
AI may:
scaffold learning
improve access
support communication
accelerate drafting
reduce some barriers
The strategic challenge is not the existence of AI itself, but:
distinguishing support from substitution.
Several sector guides explicitly encourage incorporating GenAI into assessment in ways that promote critical, ethical and disciplinary use, rather than prohibiting tools outright (Australasian Academic Integrity Network, 2023; Tertiary Education Quality and Standards Agency, 2023; Lodge et al., 2025).
6. Emerging Adaptation Patterns
Pattern 1 — Conversational Capability Verification
Interest is growing in the use of:
short professional conversations
structured reflective kōrero
conversational confirmation
oral verification processes
explanation-based assessment
particularly within:
online delivery
vocational education
work-based learning
capability-focused programmes
These approaches are not necessarily new within Aotearoa New Zealand contexts. Many providers already use forms of professional discussion and reflective kōrero as part of assessment and learner support practice.
AI-assisted content generation increases the strategic importance of these approaches as ways to strengthen confidence in learner understanding, judgement, and applied capability.
The shift is subtle but important:
from verifying the artefact alone toward verifying the learner’s ability to meaningfully explain, apply, and stand behind the work produced.
Assessment reform work now advocates for more performative, process-visible and socially moderated assessments, including professional discussions and live demonstrations, as more resilient ways to assure learning in AI-rich environments (Lodge et al., 2025).
Pattern 2 — Process Visibility
Movement toward:
staged submissions
checkpoints
rationale explanation
reflective commentary
iterative evidence collection
Pattern 3 — Applied Demonstration
Increasing emphasis on:
contextual application
real-world tasks
live demonstrations
observable performance evidence
Pattern 4 — AI-Integrated Assessment
Some educators are beginning to incorporate AI use directly into assessment itself.
Sector guidance similarly recommends explicitly integrating GenAI into assessment tasks so that students learn to use these tools responsibly while being assessed on judgement, critique and synthesis rather than unaided production (Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2023).
The focus shifts toward:
judgement
critique
synthesis
interpretation
responsible use
rather than unaided production alone.
7. Strategic Implications
Short-Term
Institutions are likely to continue operating in mixed-mode uncertainty.
Variation between providers will likely widen.
Medium-Term
Pressure is likely to increase for:
clearer operational guidance
moderation adaptation
capability verification approaches
assessment redesign capability
staff development
Large-scale providers are unlikely to adopt high-intensity verification universally. More probable are risk-aware approaches where stronger verification is selectively applied where capability stakes or evidence ambiguity are highest.
National and institutional papers increasingly converge on three broad pathways: programme-level assessment reform, unit-level assurance of learning, and hybrid approaches that mix secure and open assessments (Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2023).
Long-Term
The sector may gradually shift from:
artefact verificationtoward: capability verification.
This may represent a significant structural transition rather than a temporary adjustment.
8. Intervention Options
| Option | Description | Strength | Risk |
|---|---|---|---|
| Do nothing / minor adjustments | Retain current settings with limited local tweaks | low immediate disruption | unstable under rising scrutiny |
| Detection-first | Continue enforcement-heavy approach | politically simple | weak long-term stability |
| Local experimentation | Allow provider-level adaptation | innovation potential | fragmentation |
| National guidance framework | Shared operational principles | greater coherence | slower development |
| Capability verification infrastructure | Build stronger verification approaches | future-aligned | implementation complexity |
| AI-integrated redesign | Accept AI presence and redesign accordingly | operational realism | uneven readiness |
Variants of these options are already visible in Australasian guidance, which contrasts detection-heavy responses with structural assessment reform and programmes that explicitly integrate GenAI into authentic tasks (Lodge et al., 2023; Lodge et al., 2025; Tertiary Education Quality and Standards Agency, 2025).
9. Emerging Directional Principles
These are not prescriptions.
They are emerging strategic principles.
Principle 1
Shift focus from:
“Was AI used?”
toward:
“What capability can be confidently verified?”
This reframing is consistent with emerging international guidance, which argues that the central question is no longer tool use per se, but whether assessment systems can still provide reliable assurance of learning (Cardona et al., 2023; Lodge et al., 2025).
Principle 2
Support institutions in building:
assessment redesign capability
oral verification literacy
AI-integrated pedagogy
moderation adaptation capability
Principle 3
Treat AI adaptation as:
operational transformation
rather than:
temporary disruption.
Principle 4
Preserve institutional flexibility while improving coherence.
Avoid premature over-standardisation.
10. Watchlist
Signals Worth Monitoring
employer trust shifts
moderation disputes
learner dependency patterns
AI-generated evidence escalation
oral verification uptake
provider capability gaps
assessment workload pressure
capability verification experimentation
11. Conclusion
The central issue is no longer whether AI is entering tertiary learning systems.
It already has.
The strategic question is now:
how capability, trust, and legitimacy are maintained under conditions where traditional assessment assumptions no longer reliably hold.
The most likely future is not the disappearance of assessment, but the gradual evolution of more adaptive, multi-modal, and verification-aware capability systems.
Assessment reform initiatives across multiple jurisdictions already point toward multi-modal, programme-level assurance approaches that treat GenAI as a permanent feature of the landscape rather than a temporary disruption (Lodge et al., 2025; Ross et al., 2020; The University of Sydney, 2024).
Institutions that adapt earlier may gain significant long-term advantages in:
learner trust
workforce confidence
moderation resilience
delivery flexibility
capability visibility
institutional credibility
Closing Observation
The challenge emerging under AI conditions is not simply technological.
It is epistemic.
Educational systems are increasingly being asked:
how they know capability is genuinely present.
That question may become one of the defining tertiary challenges of the AI era.
References
Australasian Academic Integrity Network. (2023, March 1). Generative artificial intelligence guidelines. https://www.teqsa.gov.au/guides-resources/higher-education-good-practice-hub/gen-ai-knowledge-hub/gen-ai-academic-integrity-and-assessment-reform
Bridgeman, A., & Liu, D. (2025, October 2). Navigating AI in higher education: Tasks ahead for 2025 and 2026. Educational Innovation, The University of Sydney. https://educational-innovation.sydney.edu.au/teaching@sydney/navigating-ai-in-higher-education-tasks-ahead-for-2025-and-2026/
Cardona, M. A., Rodríguez, R. J., & Ishmael, K. (2023, May). Artificial intelligence and the future of teaching and learning: Insights and recommendations. U.S. Department of Education, Office of Educational Technology. https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf
Kofinas, A. K., Tsay, C. H.-H., & Pike, D. (2025). The impact of generative AI on academic integrity of authentic assessments within a higher education context. British Journal of Educational Technology. Advance online publication. https://gala.gre.ac.uk/id/eprint/50077/
Tertiary Education Quality and Standards Agency. (2023). Assessment reform for the age of artificial intelligence. https://www.teqsa.gov.au/sites/default/files/2023-09/assessment-reform-age-artificial-intelligence-discussion-paper.pdf
Lodge, J. M., Bearman, M., Dawson, P., Gniel, H., Harper, R., Liu, D., McLean, J., Ucnik, L., et al. (2025). Enacting assessment reform in a time of artificial intelligence. Tertiary Education Quality and Standards Agency. https://www.teqsa.gov.au/sites/default/files/2025-09/enacting-assessment-reform-in-a-time-of-artificial-intelligence.pdf
Ross, J., Curwood, J. S., & Bell, A. (2020). A multimodal assessment framework for higher education. E-Learning and Digital Media, 17(4), 290–306. https://www.pure.ed.ac.uk/ws/portalfiles/portal/145751051/RossEtalEDM2020AMultiModalAssessmentFramework.pdf
Tertiary Education Quality and Standards Agency. (2025). Gen AI – academic integrity and assessment reform [Knowledge hub]. https://www.teqsa.gov.au/guides-resources/higher-education-good-practice-hub/gen-ai-knowledge-hub/gen-ai-academic-integrity-and-assessment-reform
The University of Sydney. (2024, November 27). University of Sydney’s AI assessment policy: Protecting integrity and empowering students. https://www.sydney.edu.au/news-opinion/news/2024/11/27/university-of-sydney-ai-assessment-policy-protecting-integrity-and-empowering-students.html
Key Questions
What assessment assumptions is AI destabilising?
Generative AI is destabilising the assumptions that writing shows thinking, submission shows authorship, and a finished assessment artefact reliably evidences capability.
Why is this more than an academic-integrity problem?
The deeper issue is evidence trust. If institutions cannot confidently interpret what an assessment artefact proves, then moderation, credential confidence, workforce signalling, and institutional legitimacy all become harder to defend.
What is capability verification?
Capability verification strengthens confidence that a learner can explain, apply, and stand behind the work produced. It may include professional conversation, reflective kōrero, live demonstration, staged evidence, and applied performance.
Why are detection-based responses unlikely to stabilise assessment trust?
Detection tools face reliability limits, false positives, rapid model change, and inconsistent institutional use. They may support judgement, but they are unlikely to become a primary trust architecture for AI-era assessment.
What should tertiary institutions monitor next?
Institutions should monitor employer trust, moderation disputes, learner dependency patterns, assessment workload pressure, AI-generated evidence escalation, and the uptake of capability-verification practices.
Continue Exploring
Signals | Signals Archive | Subscribe to Signal Intelligence | AI Capability & Judgement | Consulting
Kia ora! Hey, I'd love to know what you think.