AI Consciousness Research
Conscium supports research into machine consciousness and how to approach it responsibly. Explore the papers and preprints published by our team or funded by Conscium, covering responsible AI consciousness research, active inference, homeostasis and affect in artificial agents, and methods for assessing consciousness risk in AI systems.
Principles for Responsible AI Consciousness Research
Butlin, P. & Lappas, T.
Sets out principles for organisations researching or building AI that could be conscious, from research objectives to knowledge sharing and communication.
Read now →Interoceptive Attention as Dynamic Homeostatic Prioritization in a Foraging Agent
Grimbly, S. J., Kuske, N., Boonstra, E. A., Bassett, B. A., van Hoof, C., Hodson, R., et al. & Shock, J. P.
Models interoceptive attention as a way for an artificial agent to dynamically prioritise its internal, homeostatic needs, explored through a foraging agent balancing competing demands.
Read now →Sophisticated Learning: A Novel Algorithm for Active Learning During Model-Based Planning
Hodson, R., Grimbly, S. J., Boonstra, E. A., Bassett, B. A., van Hoof, C., Rosman, B., et al. & Smith, R.
Introduces an algorithm that brings active learning into model-based planning, so an agent can plan ahead to seek out information that improves its own model of the world.
Read now →Assessing Consciousness Risk in Artificial Systems: A Precautionary Rubric Derived from the Spinning Wheel Theory
Hulme, D. & Griffiths, L.
Proposes a precautionary rubric, derived from the Spinning Wheel Theory, for assessing the risk that an artificial system could be conscious.
Read now →Inferring Affective Consciousness in an Artificial Agent: A Case Study
Solms, M., et al.
A case study examining whether, and on what grounds, affective consciousness could be inferred in an artificial agent.
Read now →