Computation & learning
Artificial Intelligence
Machines that predict, generate, and increasingly reason.
Modern AI made prediction and generation abundant. The open problem is no longer capability in isolation but how machine inference is composed into durable, accountable reasoning at the scale of an institution.
What it contributes
Provides the inference substrate — retrieval, extraction, prediction, and generation — that an organization orchestrates into judgment. AI is a service the reasoning layer calls, not the destination.
Foundational ideas
- Representation learning
- Turning raw signal into usable structure.
- Retrieval & grounding
- Tying generation to evidence.
- Agency & orchestration
- Composing models into workflows.
Open questions
- How should machine inference and human judgment be partitioned?
- What makes a chain of machine reasoning auditable after the fact?