The Tertiary Education Quality and Standards Agency (TEQSA) has released its long-awaited guide for alarmed academics, Assessment Reform in a Time of Artificial Intelligence.
This is a major resource, the work of Jason Lodge (Uni Queensland) and a cast of dozens. It is based on the apparent premise that as AI cannot be dismissed or destroyed, universities will have to live-with it and design assessment practise accordingly, “with full recognition of the equal imperative to promote assessment that equips students for a world where gen AI is increasingly ubiquitous.”
As such, it appears a reversal of the regulator’s response to contract cheating providers where it hoped legislation and the courts would end their business model.
The new guide builds on previous TEQSA guidance and “action plans” provided to the agency by all Australian universities in July 2024. “Since detecting gen AI use with certainty in assessments is, at this point, all but impossible, we need alternative approaches to complement academic integrity processes,” is the pitch.
And so, the paper points to three assessment reform pathways being followed:
- Assuring learning across whole degrees: “ambitious but achievable” with “multiple, interconnected, secure points of evidence that collectively provide robust assurance of student learning”
- Checking at subject/unit level: with a minimum one “secure assessment task,” “to provide evidence that students can independently demonstrate the key learning outcomes,” which aligns with traditional academic structures
- Some of both, looking at some units in overall degree assessment: “to assure evidence of progression toward program learning outcomes at key junctures in the programme.”
But what Professor Lodge and colleagues do not want is, “inequitable assessment formats that only focus on assuring learning,” such as “secure assessment formats such as invigilated time-limited tests and exams (which) “reduce assessment variety and authenticity.”