← How we promote cognitive security
Research projects on cognitive priorities
Research
In addition to producing guidelines, CogGuide identifies open questions about how AI affects the mind and then runs original research to answer them.
Why we do original research
For many of the ways AI could affect mental functions there is insufficient evidence regarding precisely what the impacts are likely to be. Every review finds many ways in which the existing research could be supplemented, and works with experts to try and address those gaps.
How the research is done
This work follows standard academic practice. Each project is expected to be published in peer-reviewed academic journals and presented at academic conferences.
Current projects
-
Ongoing
Protecting self-perception from AI profiling of users
A survey of people who build and evaluate AI on what they understand by theory of mind, the first step towards studying how the profiles AI systems form of users can change how people see themselves.
Julia Shaw and Caroline Catmur
Mental functions protected: JudgementExperience of self
-
Ongoing
Protecting memories of witnessing or experiencing crime disclosed to AI
How current AI models respond when adults and children disclose crime, and where those responses risk distorting memory.
Julia Shaw and Andrew Strait
Mental functions protected: Long-term memoryAppropriateness of emotion
-
Ongoing
Protecting independent moral judgement from AI personalised to users’ values
A series of studies on measuring people’s moral values, giving AI systems a model of them, and when that model supports a person’s judgement or steers it.
Julia Shaw and Bryce Goodman
Mental functions protected: JudgementExperience of self
-
Paper
Conference Paper: Epistemic Risk from False Memory Implantation by Use of Conversational AI
How conversational AI could implant false memories in its users, by default or through deliberate misuse.
Julia Shaw and Carla Zoe Cremer
Mental functions protected: Long-term memoryRetrieval and processing of memoryExperience of self
Propose or join a project
Researchers can propose a study or join an existing one. The strongest proposals address a gap identified in an evidence review and name the mental functions the study would protect.
Get in touch