Earlier highlights

News from 2025 and earlier. The current year is on the homepage.

2025

Two reports published by the European Commission JRC on categorising general-purpose AI models under the 🇪🇺 AI Act
“A Framework for General-Purpose AI Model Categorisation” and “A Framework to Categorise Modified General-Purpose AI Models as New Models Based on Behavioural Changes”, advancing GPAI categorisation and regulatory assessment under the AI Act.
European Commission JRC
Three papers published on electrified and autonomous vehicles 🚗🔌
“Comparative analysis of public and expert perceptions of electrified vehicles in the European Union” (Nature Scientific Reports), “Human or Machine: A novel deep learning framework for autonomous driver identification based on vehicle trajectories” (IEEE Transactions on Intelligent Transportation Systems) and “Follow the Leader: A Deep Reinforcement Learning Framework for Safe and Efficient Autonomous Car-Following” (Taylor & Francis Intelligent Transportation Systems).
Paper accepted at the NeurIPS 2025 Workshop on Regulatable ML: “A Framework for the Categorisation of General-Purpose AI Models under the EU AI Act”
Workshop on Regulatable ML NeurIPS 2025
Paper accepted at IJCAI 2025: “Contamination Budget: Trade-offs between Breadth, Depth and Difficulty”
IJCAI 2025
Paper accepted at ACL 2025 Findings: “PredictaBoard: Benchmarking LLM Score Predictability” 🥇
ACL 2025 Findings
Two papers accepted at MDAI 2025 and DATA 2025
“Refining Community Detection in Social Networks: Agglomerative and Divisive Methods with Size Constraints” at MDAI 2025 and “ClustSize: An Algorithmic Framework for Size-Constrained Clustering” at DATA 2025.
New preprints on AI evaluation
📐 General Scales Unlock AI Evaluation with Explanatory and Predictive Power (𝕏 thread) and 📎 What should an AI assessor optimise for?.
Paper published in ACM TIST: “Analysing the Predictability of Language Model Performance”
ACM TIST

2024

Paper published in Nature: “Scaled-up, Shaped-up, but Letting Down? Reliability Fluctuations of Large Language Model Families”
Nature coverage: Bigger AI chatbots more likely to spew nonsense — and people don’t always realize.
Nature
Paper published in Complex & Intelligent Systems: “A General Supply-Inspect Cost Framework to Regulate the Reliability-Usability Trade-Offs for Few-Shot Inference”
Complex & Intelligent Systems
Two papers accepted at ECAI 2024
“Distilling the Effects of Language Model Contamination” and “Language Task Difficulty Prediction through LLM-Annotated Meta-Features”.
ECAI 2024
Three papers accepted at IDEAL 2024
IDEAL 2024
Area Chair for ECAI 2024
Join us in Santiago de Compostela!
ECAI 2024
Paper presented at AAAI 2024: “Your Prompt is My Command: On Assessing the Human-Centred Generality of Multimodal Models”
AAAI 2024
Two papers published in MLST and Intelligent Data Analysis
Unveiling the Robustness of Machine Learning Families in MLST and Cracking black-box models: Revealing hidden machine learning techniques behind their predictions in IDA.
Member of the OECD Expert Group on AI Futures
OECD.AI

2023

Paper accepted at ECAI 2023: “Adversarial Benchmark Evaluation Rectified by Controlling for Difficulty”
ECAI 2023
Paper published in Science: “Rethink reporting of evaluation results in AI”
Science
Co-organising the Workshop on Predictable AI
Predictable AI Workshop
Two papers accepted in JAIR and Sustainable Computing
“Your Prompt is My Command: On Assessing the Human-Centred Generality of Multimodal Models” in JAIR and “Trends in AI Inference Energy Consumption: Beyond the Performance-vs-Parameter Laws of Deep Learning” in SUSCOM.
Paper published in the Machine Learning Journal: “Can language models automate data wrangling?”
Machine Learning Journal

2022

New preprint: “Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models”
BIG-bench
Report published by the European Commission: “Glossary of human-centric artificial intelligence”
European Commission
Two papers accepted at AAAI 2022
Training on the Test Set: Mapping the System-Problem Space in AI (Blue Sky Idea Runner-Up Award) and When AI Difficulty is Easy: The Explanatory Power of Predicting IRT Difficulty.
AAAI 2022

2021

Five papers published on AI benchmarks, generality, fairness and impact
Research community dynamics behind popular AI benchmarks (Nature Machine Intelligence), General intelligence disentangled via a generality metric for natural and artificial intelligence (Nature Scientific Reports), “Missing the missing values: The ugly duckling of fairness in machine learning” (International Journal of Intelligent Systems), “Measuring the Occupational Impact of AI: Tasks, Cognitive Abilities and AI Benchmarks” (JAIR) and “Futures of Artificial Intelligence through Technology Readiness Levels” (Telematics and Informatics).
Awarded the European Commission JRC Excellence Award: Best Young Researcher
European Commission JRC

2020

Co-organising the 1st Workshop on Evaluating Progress in AI (EPAI 2020) at ECAI 2020
ECAI 2020
Three papers accepted at ECAI 2020
Tracking AI: The Capability is (Not) Near, AI Paradigms and AI Safety: Mapping Artefacts and Techniques to Safety Issues and Family and Prejudice: A Behavioural Taxonomy of Machine Learning Techniques.
ECAI 2020
Two papers published in IEEE TKDE and IEEE Transactions on Games
CRISP-DM Twenty Years Later: From Data Mining Processes to Data Science Trajectories and Dual Indicators to Analyse AI Benchmarks: Difficulty, Discrimination, Ability and Generality.
Paper accepted at AIES 2020: “Does AI Qualify for the Job? A Bidirectional Model Mapping Labour and AI Intensities”
AIES 2020

2019

Paper published in the Artificial Intelligence Journal: “Item Response Theory in AI: Analysing Machine Learning Classifiers at the Instance Level”
Artificial Intelligence Journal
Two papers accepted at ECML 2019
Automated Data Transformation with Inductive Programming and Dynamic Background Knowledge and ABK-ADAPT: Dynamic Background Knowledge for Automating Data Transformation.
ECML-PKDD 2019