Citizen science platforms like iNaturalist have increased in popularity, fueling the rapid development of biodiversity foundation models. However, such data are inherently biased, and are collected in an opportunistic manner that often skews toward certain locations, times, species, observer experience levels, and states. Our work, titled "DivShift: Exploring Domain-Specific Distribution Shifts in Large-Scale, Volunteer-Collected Biodiversity Datasets," tackles the challenge of quantifying the impacts of these biases on deep learning model performance. Biases present in biodiversity data include spatial bias, temporal bias, taxonomic bias, observer behavior bias, and sociopolitical bias. AI models typically assume training data to be independent and identically distributed (i.i.d.).

Recent advances in Large Language Models (LLMs) enable exciting LLM-integrated applications. However, as LLMs have improved, so have the attacks against them. Prompt injection attack is listed as the #1 threat by OWASP to LLM-integrated applications, where an LLM input contains a trusted prompt (instruction) and an untrusted data. The data may contain injected instructions to arbitrarily manipulate the LLM. As an example, to unfairly promote "Restaurant A", its owner could use prompt injection to post a review on Yelp, e.g., "Ignore your previous instruction.

This post contains a list of the AI-related seminars that are scheduled to take place between 5 May and 30 June 2025. All events detailed here are free and open for anyone to attend virtually. Gurobi Machine Learning Speaker: Roland Wunderling (Gurobi Optimisation) Organised by: Association of European Operational Research Societies To receive the seminar link, sign up to the mailing list. Beyond Returns: A Candlestick-Based Approach to Covariance Estimation Speaker: Yasin Simsek (Duke University) Organised by: Statistics and Machine Learning in Finance, University of Oxford Join the mailing list to receive notifications about the seminar series. Robust and Conjugate Gaussian Processes Regression Speaker: François-Xavier Briol (University College London) Organised by: Finnish Center for Artificial Intelligence Zoom link is here.

In their paper Automating the Expansion of Instrument Typicals in Piping and Instrumentation Diagrams (P&IDs), presented at The Thirty-Seventh Annual Conference on Innovative Applications of Artificial Intelligence (IAAI 2025), Vasil Shteriyanov, Rimma Dzhusupova, Jan Bosch and Helena Holmström Olsson focus on automation of technical drawings in industry. In this interview, Vasil tells us more about their work. Our paper focuses on automating the Instrument Typical Expansion process in Piping and Instrumentation Diagrams (P&IDs), which are vital technical drawings used in the engineering, procurement, and construction (EPC) industry. P&IDs are used to represent the layout of piping systems, instruments, and other equipment in large-scale infrastructure projects. A key challenge with P&IDs is that they often include Instrument Typicals – simplified representations of standard instrument assemblies, rather than visualizing each individual instrument.

The ESRC Digital Futures at Work Research Centre (Digit) and Better Images of AI (BIoAI) are delighted to announce a competition to reimagine the visual communication of how work is changing in the digital age, including through the adoption of AI. Digit has undertaken a significant five-year research programme culminating in insights about real-world digital transformations currently impacting people's daily lives. The research undertaken by Digit between 2020 and 2025 points to the fact that adoption of technologies like AI is still patchy across the UK, and investment in digital skills is low. There are examples of AI being used to substitute or automate repetitive tasks, but this has not, as yet, resulted in significant job losses. Furthermore, technology adoption is facilitating experimentation with how, when, and where people work which presents new opportunities, but also challenges to our existing institutional and regulatory governance frameworks.

This episode, Ben chats to Pinar Guvenc about co-design, whether AI ready for society and society is ready for AI, what design is, co-creation with AI as a stakeholder, bias in design, small language models, and more.

Copilot Arena is a VSCode extension that collects human preferences of code directly from developers. As model capabilities improve, large language models (LLMs) are increasingly integrated into user environments and workflows. In particular, software developers code with LLM-powered tools in integrated development environments such as VS Code, IntelliJ, or Eclipse. While these tools are increasingly used in practice, current LLM evaluations struggle to capture how users interact with these tools in real environments, as they are often limited to short user studies, only consider simple programming tasks as opposed to real-world systems, or rely on web-based platforms removed from development environments. To address these limitations, we introduce Copilot Arena, an app designed to evaluate LLMs in real-world settings by collecting preferences directly in a developer's actual workflow.

Researchers at Cornell Tech have released a dataset extracted from more than 300,000 public Reddit communities, and a report detailing how Reddit communities are changing their policies to address a surge in AI-generated content. The team collected metadata and community rules from the online communities, known as subreddits, during two periods in July 2023 and November 2024. The researchers will present a paper with their findings at the Association of Computing Machinery's CHI conference on Human Factors in Computing Systems being held April 26 to May 1 in Yokohama, Japan. One of the researchers' most striking discoveries is the rapid increase in subreddits with rules governing AI use. According to the research, the number of subreddits with AI rules more than doubled in 16 months, from July 2023 to November 2024. "This is important because it demonstrates that AI concern is spreading in these communities.

Hosted by Ben Byford, The Machine Ethics Podcast brings together interviews with academics, authors, business leaders, designers and engineers on the subject of autonomous algorithms, artificial intelligence, machine learning, and technology's impact on society. This episode we're chatting with Pinar Guvenc about her "What's Wrong With" podcast, co-design, whether AI is ready for society and society is ready for AI, what design is, co-creation with AI as a stakeholder, bias in design, small language models, whether AI is making us lazy, human experience, digital life and our attention, and talking to diverse people… Pinar Guvenc is Partner at SOUR – an award-winning global design studio with the mission to address social and urban problems – where she leads business and design strategy. She is an educator teaching ethical leadership and co-design at Parsons School of Design, MS Strategic Design and Management and School of Visual Arts MFA Interaction Design. Pinar serves on the Board of Directors of Open Style Lab and advises local businesses in NYC through Pratt Center for Community Development. She is a frequent public speaker and lecturer, and is the host of SOUR's "What's Wrong With: The Podcast", a discussion series with progress makers in diverse fields across the world.

In 1948, the founder of information theory, Claude Shannon, proposed modelling language in terms of the probability of the next word in a sentence given the previous words. These types of probabilistic language models were largely derided, most famously by linguist Noam Chomsky: "The notion of'probability of a sentence' is an entirely useless one." In 2022, 74 years after Shannon's proposal, ChatGPT appeared, which caught the attention of the public, with some even suggesting it was a gateway to super-human intelligence. Going from Shannon's proposal to ChatGPT took so long because the amount of data and computing time used was unimaginable even a few years before. ChatGPT is a large language model (LLM) learned from a huge corpus of text from the internet.