The evidence base for experiential and project-based learning is unusually consistent by the standards of educational research. Study after study across multiple decades and contexts shows that students learn more durably when they apply knowledge to authentic problems, receive feedback on their performance in real-world-adjacent tasks, and have the opportunity to reflect on the gap between what they attempted and what they achieved. This is not a controversial finding in the learning sciences — it's one of the most replicated results in the field. What's controversial is the implication: if experiential learning is pedagogically superior for long-term knowledge retention and transfer, why does classroom-based lecture-and-assessment remain the dominant delivery model in higher education and upper secondary schooling?
The answer is facilitation cost. Experiential learning requires one-to-one or small-group feedback at critical points in the learning cycle — when a student is working through a problem and hits a decision point, when an artefact or performance needs evaluating against professional standards, when a student is trying to make sense of a setback in a work-integrated learning placement. That feedback needs to come from someone with professional or domain knowledge, who can recognise the specific gap between what the student attempted and what a competent practitioner would have done. Scaling that requires either a very high ratio of expert facilitators to students, or a technology that can approximate the expert facilitation role at the moments where it matters. Until recently, neither was economically viable at the scale of a university cohort or a large-scale industry training programme.
AI changes the facilitation economics in a specific way. It doesn't replicate the expert facilitator — a student navigating a complex work-integrated learning placement where the project requires genuine professional judgment is not well served by a chatbot. What AI can do is handle the scaffolding overhead that currently falls to human facilitators: the check-in that confirms a student understands the task brief before they invest three hours in the wrong direction, the guided reflection prompt that helps a student identify what to improve before the expert supervisor review, the synthesis of a student's learning journal into the key themes that an assessor needs to evaluate. These are not the intellectually demanding parts of experiential facilitation. They are the high-volume, time-consuming parts that currently prevent facilitators from spending their time on the high-value interactions.
Practera, which we backed in 2025, has built a platform for exactly this model: AI-managed experiential learning at scale, where the AI layer handles programme management, student check-ins, submission scaffolding, and initial feedback synthesis, while human industry mentors and academic supervisors focus their limited attention on the substantive professional judgment conversations. The programmes running on Practera include work-integrated learning cohorts, industry project programmes, and internship management at Australian and NZ universities. The feedback from program coordinators is consistent: AI-managed check-ins significantly reduce the administrative burden that was previously preventing facilitators from having the quality feedback conversations that make experiential learning effective.
The structural implication for higher education is significant. Work-integrated learning and experiential programmes have historically been the exclusive domain of well-resourced institutions with strong industry relationships and enough staff to manage the facilitation complexity. If AI can reduce the facilitation overhead sufficiently, experiential learning programmes become accessible to institutions that currently couldn't run them — smaller regional universities, community colleges, vocational training providers. That expands access to a pedagogy with genuinely strong learning outcomes, which matters for equity as much as for institutional competitiveness. We think this is one of the more important applications of AI in education, even though it doesn't carry the same conversational currency as AI tutoring or AI-generated content.