Two linked days
The strand runs on Tuesday 7 July and Wednesday 8 July, with morning and afternoon sessions each day.
Back to Teaching
Exploring the relationship of AI and human intelligence
Dominik taught a four-part lecture-discussion strand during AI Exploration Week at Oxford: from historical roots to the agentic present. The sessions move from reasoning and computing, through Large Language Models, to agentic AI systems and the changing role of expertise in AI-mediated work.
AI Exploration Week, Oxford, 6-10 July 2026 · The Walter Room, Kellogg College
Kellogg College
All four sessions are listed for The Walter Room at Kellogg College, 62 Banbury Road.
One argument
The sessions can be taken separately, but they work best as a sequence about AI, meaning, tools, and judgement.
Sessions
Part 1: AI as Reasoning and Computing: History of a Joint Idea
Tuesday 7 July 2026 · 10:00-12:00
A historically grounded introduction to AI's development, from symbolic methods to deep learning, and to the older arguments about minds, machines, language, perception, and computation that still matter now.
Takeaway. A map of pre-LLM AI and why the current moment is both new and familiar.
Part 2: Large Language Models: Development and Key Concepts
Tuesday 7 July 2026 · 14:00-16:00
A practical history of LLMs from embeddings, attention, transformers, scaling, and instruction tuning to chat interfaces, multimodal systems, reasoning models, and agentic tool use.
Takeaway. A working vocabulary for understanding what LLMs are doing and why they behave as they do.
Part 3: Agent Architecture and Context Engineering: From Chatbots to AI Systems
Wednesday 8 July 2026 · 10:00-12:00
An introduction to the shift from chatbots to agentic systems, including architecture, memory, tools, planning, orchestration, and context engineering for dependable AI workflows.
Takeaway. A clearer view of what changes when AI can work with tools, files, memory, and plans.
Part 4: Expertise and AI: Past and Future
Wednesday 8 July 2026 · 14:00-16:00
A session on how AI changes where expertise is needed, why judgement and context become more important, and what earlier technologies can teach us about shifts in skill and responsibility.
Takeaway. A more precise account of what AI makes easier, what it makes harder, and what expertise is still for.
About the sessions
AI is usually presented as a sudden arrival. This series gives participants a historically grounded way of understanding why today's AI looks the way it does, what is genuinely new, what is recurring, and what kinds of human expertise become more important rather than less.
Who this is for
These sessions are for Oxford academics, researchers, professional staff, and knowledge workers who want more than a tool tour. They are for people who want to understand the ideas behind current AI systems well enough to use them critically, creatively, and responsibly.
You do not need a technical background. You do need curiosity about how the current AI moment came to be, why Large Language Models behave as they do, and what kinds of judgement become more important when AI becomes part of everyday work.