Synthetic Organisations in Practice: Investigating the Behavioural Impacts of AI-Personae Systems on Knowledge Work Award Details Award Type:AcceleratorCommissioning Fund Theme:Use of AI in behavioural research Lead Applicant:Dr Shiyan zhangAmount Awarded:£50,750 Full Economic CostsAdministering Institution:University of ManchesterStart Date:1st April 2026Duration:18 monthsProject Partners: Stevens Institute of TechnologyMore Music Research Summary This project investigates how synthetic personae — conversational agents enabled by generative AI technologies such as chatGPT and Claude — influence behaviour, decision-making, and organisational practices in real-world settings. Synthetic personae are AI-generated agents with distinct roles, communication styles, and knowledge domains that simulate collaborative dialogue. Instead of treating AI as a passive assistant, this research examines AI as an active participant in group work—potentially reshaping how people interact, share responsibility, and reach decisions in the workplace. Aligned with the aims of Behavioural Research UK (BR-UK), this project explores how people adapt to complex socio-technical systems, how organisational design impacts behaviour, and how behaviourally informed interventions can support responsible technology use. We focus on behavioural responses to AI systems embedded in everyday work—especially trust, delegation, cognitive framing, and interpersonal dynamics. Building on a prototype being developed with support from an AHRC IAA collaboration, the project will run a series of co-designed behavioural experiments with partners in the creative and policy sectors. Participants will complete knowledge-intensive tasks (e.g., planning, evaluation, synthesis) using AI collectives, while we analyse behavioural patterns and communication strategies. The project will generate tools, guidance, and insight to help organisations design for better outcomes when integrating AI into collaborative workflows. Over 18 months, we aim to deliver practical recommendations, visualisations, and open-access toolkits to help organisations navigate the behavioural impacts of AI integration. Our findings will be shared widely and will lay the groundwork for a larger programme of behavioural research on human-AI collaboration in the GenAI era. Expected Deliverables, Outputs and Outcomes OutputsEnhanced, deployable synthetic personae prototype with comprehensive documentation Detailed organisational case studies documenting behavioural interactions Practical visual tools including infographics and behavioural typology maps Behavioural insight reports tailored to policy makers and organisational stakeholders Two interactive dissemination workshops and open-access toolkit Dataset and annotated transcripts for secondary research use Minimum two academic manuscript submissions.OutcomesUnderstanding behavioural change – Analysis of case studies, interaction transcripts, and insight reports will reveal how synthetic personae influence trust, delegation, and decision-making in realistic organisational tasks. Evidence-based design guidance – Toolkits, typology maps, and policy-focused briefs will provide actionable, empirically grounded recommendations for integrating AI agents into collaborative workflows. Behavioural interventions – Workshops and behavioural mapping will identify interventions to improve trust, delegation, and decision-making, embedded in the open-access toolkit for wide adoption. Foundation for future research – Publications, methodological documentation, and datasets will establish a replicable evidence base for advancing studies on synthetic organisational roles. Organisational readiness – Stakeholder engagement and toolkit use will equip organisations with practical strategies for responsible adoption of synthetic personae. Research Team Name Organisation Role Dr. Marina Bazhydai Lancaster University Advisor Prof. Jeffrey V. Nickerson Stevens Institute of Technology Advisor Brian Green Lancaster University Technical Consultant David McBride Technician This article was published on Wednesday 8 July 2026