Human-AI Interaction в рабочих процессах

IPIIR's line on human–AI interaction at work runs empirical studies of how employees and teams fold AI tools into everyday work. The focus is observed behaviour: when a person accepts a model's recommendation, when they push back, when they trust it blindly and when their distrust is well founded. Methods come from the sociology of work, usability research and telemetry analysis.

A trust interface is how a tool signals how far its output can be relied on: an explanation of the decision, a confidence estimate, an explicit handover to a human. The line tests which of those elements genuinely change behaviour and which stay decorative, and turns the findings into design guidance for workplace tools.

A survey records what an employee believes about working with AI; telemetry records what they actually do. The gap between declared and observed use is itself a research subject: the Institute measures adoption effects from behaviour rather than from self-assessment, and reads both sources against each other within a single study.

Fieldwork proceeds only with the informed consent of participants: an employee knows what is collected and why. Findings are de-identified, and no individual worker or workstation is identifiable in publications. The ethics of observing AI-assisted work is a research topic of the line in its own right, with methodological guidance published openly.