AI Consulting for Education

As a policy and assessment-design question first, and a detection question a distant second, because the detection tools are not reliable enough to hang a misconduct finding on. So-called AI-detectors produce both false positives and false negatives, and false positives are the dangerous ones — accusing a student of misconduct on the strength of a probability score is a serious fairness problem, and several institutions have pulled back from relying on them for exactly that reason. The more durable response is to redesign assessment so that it either credits appropriate AI use openly or is genuinely resistant to it: oral defences, in-class and supervised work, process portfolios that show drafting, personalised and applied tasks, and clear disclosure requirements. For higher education this sits under the sector’s academic integrity expectations, and TEQSA has been active on generative AI and assessment. We help you think through the operational and technology side of that; the academic policy itself belongs with your teaching and learning leadership, not a technology consultant.

They are real, they are layered, and they are the first thing to establish rather than the last. Depending on your institution you may be subject to the federal Privacy Act 1988 and the Australian Privacy Principles, and separately to state and territory information-privacy laws — for example the Privacy and Data Protection Act 2014 in Victoria, the Privacy and Personal Information Protection Act 1998 in New South Wales, and the equivalents in other jurisdictions — which can apply to public schools and universities differently from private providers. On top of that, student records frequently include sensitive information, and where you deal with children there are heightened expectations around consent and data handling. What that means in practice for an AI system is that where student data physically goes, whether a vendor retains or trains on it, and which jurisdiction it sits in are not technical footnotes, they are the core of the decision. We establish those facts so your privacy officer and legal counsel can advise on them — we do not certify your privacy compliance, and no technology consultant should claim to.

Meaningfully, and it changes both the constraints and the opportunities. As a registered training organisation you sit under the VET regulator, ASQA, and the Standards for RTOs, which put specific weight on the integrity and validity of assessment and on you holding evidence that a student is genuinely competent. That makes generative AI in assessment a direct compliance concern, not just an academic-integrity one: if AI use undermines your ability to demonstrate valid, authentic assessment, that is an audit exposure. On the other side, RTOs carry a heavy administrative and compliance-reporting load — enrolment, AVETMISS data, student communications, trainer resource development — and that admin is often exactly where AI pays off soonest with the least risk. We help RTOs separate the high-risk assessment questions, which need careful policy work with your compliance function, from the lower-risk administrative wins that can proceed sooner. We do not represent ourselves as able to sign off your ASQA compliance; that remains yours.

Overwhelmingly in administration and support rather than in anything that touches assessment or academic judgement, which is where the risk concentrates. The reliable wins look similar across schools, universities and RTOs: student and parent enquiry triage, where a large volume of repetitive “when, where, how do I” questions can be handled or routed with a human on anything sensitive; enrolment and admissions administration; drafting and formatting learning resources for an educator to review and own; timetabling and operational scheduling; first-draft feedback that a teacher edits and stands behind rather than sends unread; and accessibility support such as captioning, transcription and reading-level adaptation, which genuinely widens access. None of that replaces the educator or the academic decision, and the ones that work keep a human firmly accountable for anything that affects a student’s record, mark or wellbeing. The glamorous “AI tutor” gets the headlines; the unglamorous admin is where the hours and the safe value actually are.

It can be, but it demands more care than an adult, business context, and “more care” has to be designed in rather than assumed. Data about children attracts heightened privacy expectations and, in many settings, specific duty-of-care and child-safety obligations that sit alongside privacy law. The practical implications for an AI system are concrete: minimise the student data that goes anywhere near a model in the first place; know exactly where it goes and whether a vendor retains or trains on it; keep a human in the loop on anything affecting a child’s record, progression or wellbeing; and be able to explain to a parent, in plain terms, what happens to their child’s information. We help you map that data path and design the system conservatively around it. What we do not do is make the child-safety or duty-of-care determination for you — that sits with your institution, your child-safety framework and your legal advisers, and any consultant who waves it away is one to avoid.

It starts with a free initial consultation about where the administrative load actually is and what your teaching, compliance and privacy leaders are worried about. If there is a real opportunity, the usual next step is the AI Opportunity Audit at around $3,000 — a written map of your workflows, the systems and student data behind them, the privacy, academic-integrity and regulatory constraints attached to each, and a ranked shortlist that deliberately separates the safe administrative wins from the high-risk assessment questions. You keep it whichever way you go. Where it makes sense we then build and ship the first low-risk workflow into production rather than handing over recommendations, with human oversight designed in. The audit generally earns its fee at around 20 or more staff. Throughout, we are explicit that we establish facts and design systems; we do not certify your compliance with privacy law, TEQSA or ASQA requirements — those determinations remain with your institution and its advisers.