Dr Zarinah Agnew
Dr Zarinah Agnew is Research Director at the Collective Intelligence Project (CIP), a non-profit that builds shared infrastructure for democratic governance of AI. Their work develops methods for eliciting, measuring, and representing public input in how AI systems are built and governed.
During the programme, Agnew will extend CIP's Digital Twin Evaluation Framework (DTEF) to study representational metacognition: whether AI models, and the groups relying on them, can tell where their representation of people fails.
The work has three strands:
- building new digital twin evaluations using fresh Global Dialogues rounds;
- measuring how representational accuracy shifts as public views change over time;
- and testing whether frontier models can predict their own accuracy in modelling specific groups.
Alongside this measurement work, Agnew will develop a normative account of when it is legitimate to delegate authority to an AI proxy, working with the Institute’s philosophers of AI.
At CIP, Agnew leads Global Dialogues, a longitudinal programme gathering cross-cultural data on public attitudes towards AI from participants in more than 80 countries, and works on models for ensuring that global public perspectives are represented in agentic systems.
Agnew trained as a neuroscientist before moving into AI governance research and holds a PhD in human neuroscience. They held academic research posts at University College London and the University of California, San Francisco for more than a decade, and have published 26 peer reviewed papers.
Agnew also runs Social Science Observatory, a non-profit exploring the science of human colectivity, and teaches at London College of Political Technology. Previously, they ran a science hotel and were an Aspen Foresight Fellow and a science consultant.
Further information
Contact
Email address: aiethicsafp@philosophy.ox.ac.uk