What would make the UK a better place to live? People across the UK are being urged to share their vision for how their community and country's future should look, as part of a major new research project. The National Conversation is being launched with voice notes submitted by high-profile figures, including former footballer Gary Lineker, Chief Rabbi Sir Ephraim Mirvis, and broadcaster Mariella Frostrup. Participants will be asked to complete a survey carried out by researchers from the University of Oxford and leave a 60-second voice note. AI models will then be used to analyse thousands of responses to map what could bring us together.

Claire chatted to Chenying Liu from University of Oxford about how a robot's physical form can actively contribute to sensing, processing, decision-making, and movement. Chenying Liu is a Junior Research Fellow and an Associate Member of Faculty in the Department of Engineering Science at the University of Oxford. She leads an independent research programme focused on embodied physical intelligence, exploring how robot design can integrate geometry, materials, and control to enhance autonomy and robustness. Her work aims to develop more efficient and resilient robotic systems by embedding intelligence directly into their physical structures. Robot Talk is a weekly podcast that explores the exciting world of robotics, artificial intelligence and autonomous machines.

Bayesian Optimal Experimental Design (BOED) provides a rigorous framework for decision-making tasks in which data acquisition is often the critical bottleneck, especially in resource-constrained settings. Traditionally, BOED typically selects designs by maximizing expected information gain (EIG), commonly defined through the Kullback-Leibler (KL) divergence. However, classical evaluation of EIG often involves challenging nested expectations, and even advanced variational methods leave the underlying log-density-ratio objective unchanged. As a result, support mismatch, tail underestimation, and rare-event sensitivity remain intrinsic concerns for KL-based BOED. To address these fundamental bottlenecks, we introduce an IPM-based BOED framework that replaces density-based divergences with integral probability metrics (IPMs), including the Wasserstein distance, Maximum Mean Discrepancy, and Energy Distance, resulting in a highly flexible plug-and-play BOED framework. We establish theoretical guarantees showing that IPM-based utilities provide stronger geometry-aware stability under surrogate-model error and prior misspecification than classical EIG-based utilities. We also validate the proposed framework empirically, demonstrating that IPM-based designs yield highly concentrated credible sets. Furthermore, by extending the same sample-based BOED template in a plug-and-play manner to geometry-aware discrepancies beyond the IPM class, illustrated by a neural optimal transport estimator, we achieve accurate optimal designs in high-dimensional settings where conventional nested Monte Carlo estimators and advanced variational methods fail.

Chaos arises in many complex dynamical systems, from weather to power grids, but is difficult to accurately model using data-driven emulators, including neural operator architectures. For chaotic systems, the inherent sensitivity to initial conditions makes exact long-term forecasts theoretically infeasible, meaning that traditional squared-error losses often fail when trained on noisy data. Recent work has focused on training emulators to match the statistical properties of chaotic attractors by introducing regularization based on handcrafted local features and summary statistics, as well as learned statistics extracted from a diverse dataset of trajectories. In this work, we propose a family of adversarial optimal transport objectives that jointly learn high-quality summary statistics and a physically consistent emulator. We theoretically analyze and experimentally validate a Sinkhorn divergence formulation (2-Wasserstein) and a WGAN-style dual formulation (1-Wasserstein). Our experiments across a variety of chaotic systems, including systems with high-dimensional chaotic attractors, show that emulators trained with our approach exhibit significantly improved long-term statistical fidelity.

What Will It Take to Get A.I. Out of Schools? The tech world assumes that A.I.-aided education is necessary and inevitable. A growing number of parents, educators, and cognitive scientists say the opposite. I don't like A.I., and I am raising my children not to like it. I've been telling them for years now that chatbots are manipulative and dangerous, that A.I.-image generators are loosening our collective grip on reality, that large language models are built atop industrial-scale intellectual-property theft. At times, I find myself speaking with my kids about A.I. in the same terms that we might discuss a creepy neighbor who lives down the block: avoid eye contact, cross the street when you walk past his house, and, when in doubt, call on a trusted adult. Yes, I, too, have suspected that the creepy neighbor walks on cloven hooves inside his Yeezy Boosts, but he probably isn't going anywhere--in fact, he keeps buying up properties around town--so just try your best not to engage. Somehow, I was not prepared for the creepy neighbor to start hanging around my kids' schools; somehow, I thought we had until high school.

Alan Winfield is Professor of Robot Ethics at the University of the West of England (UWE), Visiting Professor at the University of York, and Associate Fellow of the Cambridge Centre for the Future of Intelligence. Alan co-founded the Bristol Robotics Laboratory, where his research is focussed on the science, engineering and ethics of cognitive robotics. Alan is an advocate for robot ethics; he chairs the advisory board of the Responsible Technology Institute at the University of Oxford and has co-drafted new standards on ethical risk assessment and transparency. Robot Talk is a weekly podcast that explores the exciting world of robotics, artificial intelligence and autonomous machines. Robot Talk is a weekly podcast that explores the exciting world of robotics, artificial intelligence and autonomous machines.

This post contains a list of the AI-related seminars that are scheduled to take place between 2 April and 31 May 2026. All events detailed here are free and open for anyone to attend virtually. What Do Our Benchmarks Actually Measure? Vukosi Marivate (University of Pretoria) University of Michigan Zoom link is here . Optimization Over Trained Neural Networks: What, Why, and How? Thiago Serra Azevedo Silva (University of Iowa) Association of European Operational Research Societies To receive the seminar link, sign up to the mailing list .

Conformal risk control (CRC) provides distribution-free guarantees for controlling the expected loss at a user-specified level. Existing theory typically assumes that the loss decreases monotonically with a tuning parameter that governs the size of the prediction set. However, this assumption is often violated in practice, where losses may behave non-monotonically due to competing objectives such as coverage and efficiency. In this paper, we study CRC under non-monotone loss functions when the tuning parameter is selected from a finite grid, a setting commonly arising in thresholding and discretized decision rules. Revisiting a known counterexample, we show that the validity of CRC without monotonicity depends critically on the relationship between the calibration sample size and the grid resolution. In particular, reliable risk control can still be achieved when the calibration sample is sufficiently large relative to the grid size. We establish a finite-sample guarantee for bounded losses over a grid of size $m$, showing that the excess risk above the target level $α$ scales on the order of $\sqrt{\log(m)/n}$, where $n$ is the calibration sample size. A matching lower bound demonstrates that this rate is minimax optimal. We also derive refined guarantees under additional structural conditions, including Lipschitz continuity and monotonicity, and extend the analysis to settings with distribution shift via importance weighting. Numerical experiments on synthetic multilabel classification and real object detection data illustrate the practical implications of non-monotonicity. Methods that explicitly account for finite-sample uncertainty achieve more stable risk control than approaches based on monotonicity transformations, while maintaining competitive prediction set sizes.

Digital Health Technologies (DHTs) including consumer wearable devices and digital health applications offer an opportunity for continuous, large-scale data collection. Wearables give insight into physiological biomarkers that help us understand the human body, through passive data collection. Such data can be collected at a regularity that would be impossible otherwise. Digital health applications provide the chance to collect diverse types of data from clinically validated surveys, GPS, and contextual inputs. This combination has the ability to make profound advances in our understanding of the factors that affect individuals on a personal and population level [Grace et al., 2025]. One of these factors is the menstrual cycle. Particularly because of its inter-individual variability, studying it requires large sample sizes, and to truly grasp its effects on the human body, it needs to be observed on a near-daily scale [Bull et al., 2019].

Quantum computers could cause a global security crisis that makes the once-feared millennium bug, or Y2K, look quaint. This infamous computer risk was averted through the persistent behind-the-scenes work of engineers across the world, but whether the new threat will be tackled similarly is an urgent yet unresolved question. Most digital communications and transactions are protected by cryptography based on mathematical problems that are unsolvable by conventional computers but are solvable by a sufficiently capable quantum computer. Researchers have understood this since the late 1990s, but the day when this capable-enough quantum computer comes online - or Q-Day - was thought to be very far in the future. Working quantum computers are now a reality, and recent leaps in how to use them are bringing Q-Day ever closer.