What is artificial intelligence?
Investigating the nature and limits of machine intelligence, reasoning systems, multimodal models, agents, memory, planning, and autonomous systems.
The programme takes contemporary systems seriously as technical objects rather than as metaphors. What these systems actually do — and what remains claim rather than demonstration — is itself a research question.
The stream tracks capability, reliability, and limitation across reasoning, memory, planning, and agentic behaviour, and distinguishes demonstrated performance from projected capability.
- 01
What are the demonstrated capabilities and limits of current systems?
- 02
Which claims about machine reasoning are established, and which are extrapolations?
- 03
How should capability be measured without importing human categories uncritically?