Developing reliable AI tools for healthcare

An early warning system for novel AI risks

AI for the board game Diplomacy

Benchmarking the next generation of never-ending learners

How undesired goals can arise with correct rewards

In conversation with AI: building better language models

From motor control to embodied intelligence

Discovering when an agent is present in a system

Perceiver AR: general-purpose, long-context autoregressive generation

Intuitive physics learning in a deep-learning model inspired by developmental psychology

Human-centred mechanism design with Democratic AI

BYOL-Explore: Exploration with Bootstrapped Prediction

Unlocking High-Accuracy Differentially Private Image Classification through Scale

Evaluating Multimodal Interactive Agents

Dynamic language understanding: adaptation to new knowledge in parametric and semi-parametric models

Emergent Bartering Behaviour in Multi-Agent Reinforcement Learning

A Generalist Agent

Active offline policy selection

An empirical analysis of compute-optimal large language model training

GopherCite: Teaching language models to support answers with verified quotes

Learning Robust Real-Time Cultural Transmission without Human Data

Probing Image-Language Transformers for Verb Understanding

Red Teaming Language Models with Language Models

Spurious normativity enhances learning of compliance and enforcement behavior in artificial agents

Improving language models by retrieving from trillions of tokens