Pre-arrival pilot describes varied experience of generative AI

Updated Sep 07, 2026

Rose Luckin’s 5 May 2026 Advance HE article examines generative AI responses from the Pre-arrival Academic Questionnaire pilot. It reports more than 5,500 incoming undergraduates at 15 English institutions in the first wave in September 2025, with 60.6% saying they had used generative AI. Advance HE article.

What this tells us about induction

The article describes varied uses, including exploration, explanation, writing support and drafting. It also reports differences by subject and a strong concentration of familiarity around ChatGPT among AI users. These are self-reported experiences within the pilot, not a direct assessment of students’ AI literacy or evidence representative of every incoming student.

The author argues for induction that recognises different starting points. Her recommendations should be distinguished from measured effects of a particular induction programme. Using a tool does not by itself demonstrate critical understanding; having little prior use does not establish inability to learn.

Jisc’s 17 April account explains that institutions selected different cohorts and administered questionnaires before, on or shortly after arrival. This qualifies a blanket description of every response as collected before university began. It reports local changes to induction and communications, without a causal evaluation of improved outcomes. Jisc pilot account.

The next-wave arrangements

The participation guide schedules Wave 2 for September–November 2026, with providers running their own questionnaires through Jisc Online Surveys and Advance HE providing benchmark analysis. It covers undergraduate and taught postgraduate versions. The guide assigns participating institutions as data controllers and Advance HE/Jisc as processors; it also describes linking institutional student IDs. These are the documented pilot arrangements, not a general permission to link any survey data. Participation guide, sections 6–11.

Our practical suggestion is to ask what entrants understand, what access they have and where guidance is unclear. Use appropriate examples from their courses and explain assessment expectations. If later feedback is compared with pre-arrival responses, make differences in questions, cohorts and timing visible.

Agree the purpose and handling of identifiers before collection. The governance checklist can help structure that discussion; a shared theme does not justify linking individual records or demonstrate an intervention’s effect.

Review note, 7 September 2026: verified the reported first-wave figures and planned second-wave arrangements, qualified timing and representativeness, and distinguished AI use from tested readiness or improved outcomes.