Abstract:

Background:
Laboratory demonstrators play a key role in first-year science learning but often receive limited preparation for managing classroom dynamics. In large classes, they must respond to varying levels of student preparedness and fluctuating energy, but training typically prioritises procedures rather than adaptive teaching strategies. Targeted professional development is needed to support energy regulation in laboratory settings, effective scaffolding within the Zone of Proximal Development (ZPD), and reflective practice. The use of artificial intelligence (AI) as a reflective tool can further enhance demonstrators’ capacity to evaluate and improve their teaching practice.

Aim:
To evaluate a two-stage professional development intervention strengthening demonstrators’ energy regulation, adaptive scaffolding, and reflective AI use.

Methods
The intervention was implemented in a first-year molecules, cells and genes course (1300 students; 53 demonstrators) at a large Group of Eight university in Australia. Workshop 1 used scenario-based energy rehearsals and structured reflection with Lybi, a custom AI coaching agent built on Microsoft Copilot. A smaller follow-up workshop focused on behavioural transfer to laboratory teaching.

Evaluation followed the Kirkpatrick model. Surveys assessed reaction and learning (Levels 1–2) and intended behavioural change (Level 3). An early-term student survey provided exploratory insight into perceived clarity and engagement (Level 4). Likert responses were analysed descriptively and open-ended responses thematically.

Results:
Thirty demonstrators completed the first survey (n = 30). Overall, 67% felt confident teaching their first laboratory. Additionally, 90% agreed the workshop reframed teaching as adaptive and 93% agreed rehearsals strengthened responses to classroom energy. Awareness of energy’s impact was high (90%), as was comfort adjusting teaching in the moment (83%).

AI reflection was more variable (37% useful; 34% likely future use). In the follow-up workshop, 4 participants completed the survey (n = 4); all reported strengthened energy regulation, clearer identification of students’ ZPD positioning, and intention to apply a strategy, while 50% agreed Lybi supported response planning.

Thematic responses highlighted applying energy regulation when managing disengaged or overly confident students. Student feedback (n >= 110; ongoing) was positive overall, with highest agreement for comfort asking questions and engagement (86% and 87%).

Discussion:
Preliminary findings suggest movement from reported learning toward intended behavioural change. AI uptake varied, indicating differing readiness for reflective use. Student perceptions of comfort and engagement align with training priorities; however, no causal claims are made.

Findings will inform refinement of the intervention and support broader implementation across disciplines that rely on laboratory or sessional teaching staff. They highlight the need for targeted, practice-based professional development to strengthen classroom practice and student experience. Future ethics-approved research will examine behavioural transfer and student outcomes more systematically.

Keywords:

Professional development, Laboratory education, Adaptive teaching, Energy regulation, Artificial intelligence.