We are in the era of generative AI, with large language models (LLMs) achieving outstanding results in various tasks. This new technology is here to stay and is already being embedded in our day-to-day lives. However, "with great power comes great responsibility" and researchers have the duty to shed light on all the aspects of genAI. In this talk, I will delve into LLMs' "good" and "bad" capabilities and their "ugly" consequences.
Matthew McCormack - Software Developer, National Oceanography Centre
BODC (British Oceanographic Data Centre) receives data with a diverse range of sources, formats, and content. Previously this was a heavily manual process requiring data managers to invest significant time into each dataset to standardise the format, apply controlled vocabularies, and apply QC to the data. This also involved moving the data across multiple file systems and disparate software tools.
Christina Matteotti, Head of Universities, EMEA Partnerships, Google
The Beyond Academia: AI & Data Science Applications seminar series at the University of Cambridge explores the transformative role of data and artificial intelligence in the world. Designed for postgraduate researchers, early-career academics, and students, the series bridges the gap between high-level research and real-world deployment.
Dr Francesco Paolo Casale - Human Technopole, Milan, Italy; Helmholtz Munich, Munich, Germany; Faculty of Informatics, Technical University of Munich, Garching, Germany
Talk title: AI in Human Genetics: Linking Genetic Variation to Phenotypes Across Scales
The Beyond Academia: AI & Data Science Applications seminar series at the University of Cambridge explores the transformative role of data and artificial intelligence in the world. Designed for postgraduate researchers, early-career academics, and students, the series bridges the gap between high-level research and real-world deployment.
Digital twins combine real-world data with digital representations of physical systems, but building useful digital twins involves more than simply collecting and displaying data. This talk presents my work on developing a real-time digital twin of a smart building, covering the journey from privacy-preserving sensor design and real-time event detection through to data processing and visualisation, and ultimately how these systems are evaluated with the people who use them.
LLM agents are increasingly deployed on long-horizon tasks with tool use, irreversible actions, and unpredictable feedback. Yet we have few principled ways to tell, mid-episode, whether an agent is on track or quietly failing. Most uncertainty quantification (UQ) research still centers on single-turn QA, a poor match for interactive agents. In this talk, I'll present a general formulation of agent UQ and the challenges unique to agentic settings, from choosing uncertainty estimators to modeling how uncertainty evolves over an interaction.
This free event is open only to members of the University of Cambridge (and affiliated institutes). Please be aware that we are unable to offer consultations outside clinic hours.
The division of labor enables us to build complex human and machine systems. Today, decentralized multi-agent systems with learned behavior are becoming more prevalent. At the same time, decentralization can come with a costly price of anarchy—agents learn to behave in ways that leave everyone worse off. Can we understand and guard against collectively irrational behavior? In this work, we formalize a close relationship between rationality (Pareto optimality) and learnability (uniform stability) for mixed Nash equilibria. This result can be understood by analogy with optimization.
How can we create a learning algorithm that generalises “out of distribution”, i.e. beyond the training data? What can we expect about errors when a model generalises OoD? And what should happen when an error is made? In this talk, we will discuss how OccamLab’s approach to these questions. We will discuss solutions that are prescribed by a Bayesian perspective, through work on learning invariance/equivariance properties, uncertainty quantification, and continual learning.
Jack Franklin - RSE, ICCS, University of Cambridge
Video games can be a useful medium for engaging with audiences of all ages. This can be beneficial for scientific outreach activities, to foster interest with members of the public and lower the barrier of entry to more technical topics.
