Opinion
Proudly Human, the human-certification company founded by former chief scientist (and former Monash Chancellor) Alan Finkel, has launched Assured Assessment with the University of Tasmania. The trust mark vouches that a university's graduates have personally earned their degrees, and certified universities can put it on graduation certificates as a signal to employers.
UTAS is starting with its new and redesigned courses, aiming to have every course meet the standard by the end of March. Proudly Human's role, Finkel told the ABC, is to come in once or twice a year as an auditor.
In Proudly Human's own summary, four tests must be met for every course learning outcome:
- Coverage: every course learning outcome is verified by a secured, must-pass test.
- Sufficiency: every course learning outcome is verified at the exit level of the award.
- Progression: outcomes are checked throughout the course, not just at the end.
- No exceptions: all key assessment tasks must be passed.
I came to this with my red pen uncapped, since a commercial trust mark for degrees is exactly the kind of thing assessment researchers enjoy marking down. The four tests hold up, though: they're a fair summary of where the field has landed, and the credit for that belongs jointly to UTAS and Proudly Human, who developed the standard together.
The full Assured Assessment Standard says it works with the University of Sydney's two-lane approach. It also makes room for programmatic assessment, where many low-stakes tasks build towards a high-stakes progression or course completion decision.
The "no exceptions" test is the one I'd most like the sector to take seriously. Most grades are built by adding marks together, so a brilliant take-home essay can cover for poor performance in a supervised exam. Under this standard, a task that checks something essential has to be passed, however well the student did on everything else.
Coverage and sufficiency are already required, more or less, by Threshold Standards 1.4.3 and 1.4.4. Those standards ask for assessment capable of confirming that all specified learning outcomes are achieved, and for students to have demonstrated those outcomes by the time they finish.
"Secured" and "must-pass" are also implied by the Threshold Standards. Once AI can complete an unsupervised task, that task can confirm very little on its own, and a graduate who passed on aggregate without ever demonstrating an outcome is exactly what 1.4.4 rules out.
But the thing is, if your university has a Course Standards Committee or similar, this is their job, not Proudly Human's. If your committee doesn't know that, give them a copy of the TEQSA-published Enacting assessment reform in a time of artificial intelligence by Jason Lodge and colleagues. It ties the risk generative AI poses to assurance of learning directly to Standard 1.4.4, and all three pathways it maps depend on secure tasks. It's all in there.
Over the years I've spoken with the Chairs of Academic Boards at most Australian universities, and I can tell you they know this is required. What's hard is the slow, expensive work of mapping every course and paying for the supervision that secure assessment needs.
So the question is, do we expect academic governance to do its job, or do we need (another!) external body to certify it?
AI use acknowledgement: this piece is Proudly Cyborg. It started as a LinkedIn post I wrote. Claude then drafted this longer version, which I revised. The arguments, and any errors, are mine.
Professor Phillip (Phill) Dawson researches assessment in the Faculty of Education at Monash University. He was one of the authors of Enacting assessment reform in a time of artificial intelligence.