Can AI Think Like a Scientist?
A new benchmark assesses whether artificial intelligence can tackle complex, open-ended scientific problems at an expert level.
Science
Can AI Critique AI? A New Framework for Testing Alignment
Researchers have developed a system where multiple artificial intelligences engage in structured dialogue to evaluate and refine strategies for ensuring AI safety.
Shaping Data with Physics: A New Approach to Generative AI
![The system investigates data generation under incomplete states, aiming to define the relationship between features given knowledge of other states-a process akin to reconstructing a missing piece by understanding the connections within the whole [latex] \implies [/latex] a challenge in extrapolating function from partial observation.](https://arxiv.org/html/2601.20462v1/x1.png)
A novel framework blends the principles of continuum mechanics and optimal transport to create more robust generative models, even with limited data.
Can Machines Truly Think Outside the Box?
A new evaluation metric aims to move beyond simple association and assess genuine creativity in large language models.
Robots See the World in Objects: A New Approach to Grasping and Manipulation
Researchers have developed a new framework that equips robots with a more human-like understanding of visual scenes, allowing for more adaptable and reliable manipulation skills.
Who’s Driving the Code? How Developer Roles Shape AI Tool Use
New research reveals that even with the rise of AI coding assistants, developer experience continues to dictate how these tools are integrated into daily workflows.
Beyond the Script: AI Co-Pilots for Immersive Planetarium Experiences
Researchers explore how large language models can assist human presenters in live, interactive planetarium shows, enhancing engagement and reducing cognitive strain.
The Rise of the Robot Documenter
New research reveals a growing trend of AI-generated contributions to software documentation, but raises concerns about the level of human oversight.
The AI Classroom: How Students Are Mastering the Art of Asking
New research reveals students aren’t just using ChatGPT, they’re actively learning how to interact with it, developing sophisticated strategies for leveraging its power and navigating its limitations.
Bridging Logic and Learning: A New Path to Scalable AI
Researchers have developed a novel system architecture that significantly accelerates probabilistic logical reasoning, paving the way for more efficient and adaptable artificial intelligence.