Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1549
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-16T06:54:56.184002Z
status: published
lens: p_space
sector: education
headline: As more Australian universities use AI to mark students’ work, is it creating a ‘slop-cycle’?
dek: Armin Alimardani has blue hair, is fresh faced, wears a smartwatch and describes himself as “a tech enthusiast”. He is an unlikely sentinel, perhaps, to warn that a drift towards academics using generative artificial intelligence to help mark students assignments could pose an existential threat to higher education. The Western Sydney University senior lecturer in law and technology has been studying and experimenting with AI for years – he is speaking from Palermo in Italy, where he is attending a series of uni…
gain_title: (none)
problem_title: Allowing staff to use generative AI in marking risks 'verification drift', where tired reviewers stop checking and hallucinations slip through, and may erode student trust and sessional marking work.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: Allowing staff to use generative AI in marking risks 'verification drift', where tired reviewers stop checking and hallucinations slip through, and may erode student trust and sessional marking work.
problem_evidence: verification drift | AI hallucinations slip through as facts
quick_read: Several Australian universities, including Western Sydney, Newcastle, Deakin, RMIT and Adelaide, now let staff use generative AI to help mark assessments and give feedback. Most set conditions, such as human responsibility for grades or student opt-outs. UNSW, Melbourne and Sydney refuse AI in marking, and others say they are still deciding.

The debate matters because marking is a core point of contact between students and academics, and critics fear AI marking of AI-written work could hollow out the value of a degree. The claimed benefits, such as better consistency or less workload, are described as largely untested, and how any oversight would work at scale is unresolved.
limitation: The article says the claimed benefits of AI marking are largely untested, and several universities have not decided their position, so the risks described are warnings rather than measured outcomes.
tag: Evidence-backed problem
key_points: Western Sydney, Newcastle, Deakin, RMIT and Adelaide give qualified permission for AI in marking; Deakin says staff may not use it to assign grades. | UNSW, Melbourne and Sydney draw the line at generative AI in marking, though they allow it in other parts of teaching. | Alimardani describes 'verification drift' where overworked reviewers become less vigilant after early AI outputs look fine. | A QUT email on an 'assessment uplift project' prompted staff worry about less human marking and precarious sessional jobs. | A QUT student said he would be 'incensed' to pay for a degree marked by AI.
rundown: Alimardani, a senior lecturer in law and technology at Western Sydney University, says colleagues have asked whether AI could ease grading. He calls this concerning, while acknowledging AI could act as a 'second pair of eyes' and that he is building models to detect inconsistencies in human assessment.

University policies vary. Newcastle lets students opt out, and WSU's spokesperson says marks, grades and feedback remain the responsibility of academic staff. Monash says AI is not currently used in marking, and Wollongong and James Cook responded similarly. Queensland and QUT gave less direct answers about their plans.

Staff and students raised further objections. An anonymous QUT academic questioned the message of using AI to mark when students are warned about its use in their own work, and cited environmental and ethical concerns plus doubts about reliability. Cameron said students value human feedback.
sources:
- journalism | The Guardian | https://www.theguardian.com/australia-news/2026/sep/28/university-ai-marking-academic-overworked-staff-student-shortcut | 2026-09-27
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

f812c19401d57e7e5261ce7a0823ea7cf8d98aa75e5082793e2e26aecae644c2

previous

0000000000000000000000000000000000000000000000000000000000000000