Fitting ray-traced physical models to EHT observations requires the generation of synthetic images, a task that is computationally demanding. This study leverages ALINet, a generative machine learning model, to efficiently produce radiatively inefficient accretion flow (RIAF) images as a function of the specified physical parameters. ALINet has previously been shown to be able to interpolate black hole images and their associated physical parameters after training on a computationally tractable set of library images. We utilize this model to estimate the uncertainty introduced by a number of anticipated unmodeled physical effects, including interstellar scattering and intrinsic source variability. We then use this to calibrate physical parameter estimates and their associated uncertainties from RIAF model fits to mock EHT data via a library of general relativistic magnetohydrodynamics models.

Mixture of large language model (LLMs) Agents (MoA) architectures achieve state-of-the-art performance on prominent benchmarks like AlpacaEval 2.0 by leveraging the collaboration of multiple LLMs at inference time. Despite these successes, an evaluation of the safety and reliability of MoA is missing. We present the first comprehensive study of MoA's robustness against deceptive LLM agents that deliberately provide misleading responses. We examine factors like the propagation of deceptive information, model size, and information availability, and uncover critical vulnerabilities. On AlpacaEval 2.0, the popular LLaMA 3.1-70B model achieves a length-controlled Win Rate (LC WR) of 49.2% when coupled with 3-layer MoA (6 LLM agents). However, we demonstrate that introducing only a $\textit{single}$ carefully-instructed deceptive agent into the MoA can reduce performance to 37.9%, effectively nullifying all MoA gains. On QuALITY, a multiple-choice comprehension task, the impact is also severe, with accuracy plummeting by a staggering 48.5%. Inspired in part by the historical Doge of Venice voting process, designed to minimize influence and deception, we propose a range of unsupervised defense mechanisms that recover most of the lost performance.

Apple's 3,500 Vision Pro sounds like a bargain compared to the price of a fresh, medical-grade cadaver. And some medical institutions have started practicing surgery using the spatial-computing headset, which doesn't require a physical human body. Replacing cadavers is just one example of how the Vision Pro has made its way into the medical field since it hit the market in February 2024. On January 30-31, 2025, Sharp Healthcare hosted the inaugural Spatial Computing Health Care Summit, where medical providers gathered to discuss their use of spatial computing, which embeds digital objects into a live feed of the real world. The same tech that allows people to play virtual Battleship with each other has moved into applications that include everything from training and education to full-fledged operations on human patients.

Deep learning has recently enabled the decoding of language from the neural activity of a few participants with electrodes implanted inside their brain. However, reliably decoding words from non-invasive recordings remains an open challenge. To tackle this issue, we introduce a novel deep learning pipeline to decode individual words from non-invasive electro- (EEG) and magneto-encephalography (MEG) signals. We train and evaluate our approach on an unprecedentedly large number of participants (723) exposed to five million words either written or spoken in English, French or Dutch. Our model outperforms existing methods consistently across participants, devices, languages, and tasks, and can decode words absent from the training set. Our analyses highlight the importance of the recording device and experimental protocol: MEG and reading are easier to decode than EEG and listening, respectively, and it is preferable to collect a large amount of data per participant than to repeat stimuli across a large number of participants. Furthermore, decoding performance consistently increases with the amount of (i) data used for training and (ii) data used for averaging during testing. Finally, single-word predictions show that our model effectively relies on word semantics but also captures syntactic and surface properties such as part-of-speech, word length and even individual letters, especially in the reading condition. Overall, our findings delineate the path and remaining challenges towards building non-invasive brain decoders for natural language.

The Event Horizon Telescope (EHT) provides an avenue to study black hole accretion flows on event-horizon scales. Fitting a semi-analytical model to EHT observations requires the construction of synthetic images, which is computationally expensive. This study presents an image generation tool in the form of a generative machine learning model, which extends the capabilities of a variational autoencoder. This tool can rapidly and continuously interpolate between a training set of images and can retrieve the defining parameters of those images. Trained on a set of synthetic black hole images, our tool showcases success in both interpolating black hole images and their associated physical parameters. By reducing the computational cost of generating an image, this tool facilitates parameter estimation and model validation for observations of black hole system.

Probabilistic machine learning methods are becoming increasingly powerful tools in data analysis, informing a range of critical decisions across disciplines and applications, from forecasting election results to predicting the impact of microloans on addressing poverty. This class of methods uses sophisticated concepts from probability theory to handle uncertainty in decision-making. But the math is only one piece of the puzzle in determining their accuracy and effectiveness. In a typical data analysis, researchers make many subjective choices, or potentially introduce human error, that must also be assessed in order to cultivate users' trust in the quality of decisions based on these methods. To address this issue, MIT computer scientist Tamara Broderick, associate professor in the Department of Electrical Engineering and Computer Science (EECS) and a member of the Laboratory for Information and Decision Systems (LIDS), and a team of researchers have developed a classification system--a "taxonomy of trust"--that defines where trust might break down in a data analysis and identifies strategies to strengthen trust at each step.

You wouldn't point to a banditry of chickadees -- yes, that's what they're called -- hanging out with a flock of ostriches and just say "look, a bunch of birds." It's technically true, of course, but one is an adorable passerine songbird and the other is basically a full-on velociraptor that really let itself go over the last 70 million years or so. Forget about how the ornithological eggheads parse birds little boxes, this is a distinction that even Jane Q. Public needs to be able to make at a glance because one of the two can absolutely kill you. So why are we still lumping different transactional tools under the "AI Contract Review" label? The companies themselves do it because buyers rarely cast a narrow net when researching a new product.

My father, Bill Broderick, who has died aged 80 of Covid-19, was an educationist ahead of his time in the field of computing. His vision and enthusiasm led to the first computer being installed in a British secondary school, the Royal Liberty school in Romford, Essex, where he was a maths teacher, in 1965. In a broadcast by the BBC programme Tomorrow's World from the school, Bill said: "Computers are as radical and important a keystone to our standard of living and industrial wellbeing as was the steam engine." Born in Farnborough, Kent, Bill was the only son of Ralph Broderick, an engineer, and Ida (nee Massey). He was educated at Lord Wandsworth college in Long Sutton, Hampshire, then went to Hull University to study mathematics.

This story appeared in the November 2020 issue as "Cracking the 414s." In the 1983 techno-thriller WarGames, David Lightman, played by a fresh-faced Matthew Broderick, sits in his bedroom, plunking away on a boxy computer using an 8-bit Intel processor. As text flashes across the screen, David's face lights up; he believes he's hacking into a video game company, but the unwitting teenager is actually facing off against a military supercomputer. "Shall we play a game?" the computer asks ominously. In the film, the subsequent showdown triggers a countdown to World War III.

MACHINE-LEARNING SYSTEMS USE DATA TO UNDERSTAND PATTERNS and make predictions. When the system is predicting which photos are of cats, you may not care how certain it is about its results. But if it's predicting the fastest route to the hospital, the amount of uncertainty becomes critically important. "Imagine the system tells you'Route A takes 9 minutes' and'Route B takes 10 minutes.' Route A sounds better," says Tamara Broderick, an associate professor in the Department of Electrical Engineering and Computer Science. "But now it turns out that Route A takes 9 minutes plus-or-minus 5, and Route B takes 10 minutes plus-or-minus 1. If you need a life-saving procedure in 12 minutes, suddenly your decision making really changes."