Building Reliable AI Agents: A New Functional Approach
![Across ten problem domains, aggregated scores reveal performance distinctions between agentic and baseline reaction models-specifically, [latex]agentics-agg[/latex], [latex]agentics-both[/latex], [latex]agentics-react[/latex], and [latex]baseline-react[/latex]-when evaluated using both the [latex]gemini-3-flash-previuew[/latex] and [latex]gpt-4.1[/latex] models.](https://arxiv.org/html/2603.04241v1/2603.04241v1/overall_performance_variable.png)
Researchers are introducing a framework to move beyond ad-hoc prompting and build agentic AI systems with the same guarantees of correctness and observability as traditional software.
Science
Shedding New Light on Robotic Grasping
A new dataset and synthesis strategy aims to improve robotic manipulation by systematically controlling and varying lighting conditions.
Designing the Perfect Mix: AI Tackles Solvent Discovery
A new neuro-symbolic AI system combines language models with physics-based simulations to efficiently create novel chemical formulations.
Robots Learn to Team Up: Achieving Stable Heavy-Load Transport
New research demonstrates how humanoid robots can learn to collaborate with humans during physically demanding tasks, maintaining balance and applying consistent force.
Beyond Memorization: Can AI Truly Discover New Biology?
A new benchmark challenges Large Language Models to move past recalling existing facts and demonstrate genuine knowledge discovery in the life sciences.
Smarter Cities, Shared Streets: The Rise of Human-Robot Delivery Networks
![The UrbanHuRo framework establishes a novel approach to human-robot collaboration in urban environments, fundamentally redefining interaction through the principles of shared autonomy and intuitive interfaces-a system designed to navigate the complexities of city life with a seamless blend of robotic efficiency and human intention [latex] \mathcal{UHR} = \{H, R, \mathcal{E}, \mathcal{I} \} [/latex], where <i>H</i> represents human input, <i>R</i> denotes robotic actions, [latex] \mathcal{E} [/latex] signifies the environment, and [latex] \mathcal{I} [/latex] embodies the interaction modalities.](https://arxiv.org/html/2603.03701v1/2603.03701v1/x2.png)
A new framework optimizes urban logistics by intelligently combining the strengths of human couriers and robotic vehicles for faster, more comprehensive service.
The AI That Builds AI: Experts Weigh In
A new study explores how the automation of artificial intelligence research could unlock rapid progress – and raise unprecedented challenges.
Teamwork Makes the Dream Work: Robots Learn to Collaborate with Humans
New research demonstrates a hierarchical learning framework enabling robots to adapt to human partners and dynamically coordinate during complex physical tasks.
AI’s Growing Role in Physics Education
A new analysis reveals the rapidly expanding research landscape of artificial intelligence applications in teaching and learning physics.
Seeing and Understanding: Robots Get a Boost in Social Intelligence
New research explores how a streamlined visual reasoning system can empower robots to better interpret their surroundings and human intentions.