Science

New research introduces a framework for building language models that reason about social situations more like humans do, moving past purely logical deduction.

A new benchmark reveals that current AI systems struggle not with planning how to answer questions about scientific data, but with accurately executing the necessary computations.

A new portable teleoperation system simplifies the process of collecting the diverse datasets needed to train robots for complex manipulation tasks.

New research tackles the challenge of overconfident predictions in simulation-based inference, improving the reliability of Bayesian methods.

A new platform using self-propelled robots demonstrates transitions between solid, liquid, and gaseous phases, revealing fundamental principles of active matter.

A new conversational AI system is helping physicists navigate vast internal datasets to accelerate discovery and collaboration.

New research reveals that social robots can effectively communicate with both younger and older adults using eye movements, though perceptions of this interaction differ with age.

A new framework uses category theory to formally define the dynamics of artificial chemistries, offering a powerful way to model and compare complex systems.

Researchers are pioneering a new method of robotic adaptation where robots can quickly learn to manipulate objects in unseen environments by ‘cloning’ visual characteristics from reference images.

A new paradigm shifts the operating system from managing applications to understanding user intent, powered by autonomous agents and natural language.