Methodological innovation in adult learner choice-making research: a presentation on locally deployed multi-agent AI systems

Presented virtually at the HERDSA Annual Conference 2026 (Singapore, 7 July), as part of the ongoing Learning Choice Tracker (LCT) programme.

Adult learning choices are rarely a single decision point — they’re layered, non-linear, and only partially accessible in retrospect. Semi-structured interviews help, but rely heavily on the researcher’s post-hoc reconstruction, which creates analytic lag between data collection and understanding.

This presentation reports on a local, human-governed multi-agent AI system built to close that lag: near-live transcript ingestion, AI-assisted analytic scaffolding, and choice-zone modelling, all sitting behind a researcher review and approval layer. The AI proposes; the researcher confirms — nothing becomes canonical without that step, and every proposal stays linked to its evidence.

The contribution isn’t AI substitution, it’s governed augmentation: a structured, auditable pattern for AI-assisted qualitative inquiry that keeps human judgement as the control point throughout.