Toni Jones, RMIT’s Lead, Education GenAI has authored an important article which serves as “A practical blueprint for embedding AI into your course design”. With the VET sector shortly to receive guidance on AI from ASQA, Jones encourages providers to move away from trying to “AI-proof” their assessments, to instead start to design courses that deliberately build AI-capable graduates. Her central argument is that educators should stop treating AI as an external threat to assessment integrity and instead ask how AI can be embedded in ways that build learner capability, judgement and agency.
While many worry that national training packages restrict providers in how flexible their course design can be – advice from ASQA on ‘Appropriate use of AI in VET’ makes it clear that this should not be a limiting factor where a training package is ‘pre-generative AI’.
And so for VET educators Jones’ article is a useful prompt to move beyond simply AI policies and/or ad hoc teaching experiments, and to start to think systematically about how AI should be incorporated into course design, workplace preparation and authentic assessment.
The first practical step Jones’ suggests providers focus on – is ‘backwards design’: that is, AI needs to be visible in the intended course learning outcomes, not added later as an optional activity. This step requires VET providers to consider where AI capability can be made explicit in program design, delivery plans, assessment instructions, workplace projects and, where training package rules allow, electives and contextualised tasks. The article gives examples of outcomes that require students to combine AI-generated material with human judgement, evaluate outputs for accuracy and bias, apply AI within a workflow, and reflect on ethical, social and professional implications.
The article then asks the reader to think about how to “redesign assessment with AI in mind.” Assessment is seen by Jones’ as the critical point where AI integration either succeeds or fails. Rather than trying to design tasks that students cannot use AI for, the article recommends designing assessments that require evidence of human judgement, discipline knowledge, reasoning and reflection. It uses the 4Ps framework:
- Product
- Performance
- Process, and
- Practices
The framework is used to help educators think about what learners will produce, what skills they must demonstrate, what steps they will take, and how they explain their ethical and professional decision-making. For VET, this is particularly relevant to assessment validation, assessor guides and workplace evidence, because different parts of a task may legitimately have different AI permissions.
The article also stresses that AI-integrated learning should produce artefacts that resemble real workplace outputs. Examples include AI-generated brand assets or video resumes for marketing students, AI-assisted design simulations for engineering students, AI-generated patient scenarios for health students, and AI-generated lesson plans for education students to evaluate and improve.
A particularly useful message for educators is that the actions learners take matters. If students simply prompt AI to produce an answer, they may learn little except how to obtain an output.
Jones also explains how RMIT’s GAILE GenAI Skills Continuum helps to identify the AI skills students need and which some learners may already have. That is:
“The task is to map three things: where students currently sit on the skill spectrum (introducing, refining or mastering), what skills are required to complete the task, and what skills the industry is demanding (job ads are now shockingly explicit about this).”
The article concludes with a link to download a one-page framework, developed by RMIT’s Mark Brown, with questions to help educators work out how to scaffold AI competencies into a course.
