AI Agents Learn to Test Their Own Hypotheses About How the World WorksAI agents that test their own hypotheses can adapt to new environments with fewer trials.22h ago22h ago
Neural Networks Need Kindergarten: Training AI Like Animals LearnNeural networks achieve rat-like performance in long-term decision tasks when they’re pretrained on fundamental cognitive skills.Jul 1Jul 1
Think Your Code Model Is Smart? Interactive Benchmarks Might Say OtherwiseInteractive feedback reshuffles code LLM rankings, showing static benchmarks miss key performance dynamics.Jun 25Jun 25
Learning to See Like Animals: How Small Objects in Dense Video Scenes Challenge AI VisionPooDLe enhances AI vision in real-world videos by improving small object detection.Jun 11Jun 11
Making Drugs With AI: How Machine Learning Can Design Better Antibodies Than Traditional MethodsCDS PhD grads, Prof Kyunghyun Cho, & colleagues developed an ML system that achieved up to 100x improvements in antibody binding.Jun 4A response icon1Jun 4A response icon1
In AI-Generated Content, A Trade-Off Between Quality and OriginalityNew research from CDS researchers maps the trade-off between originality and quality in LLM outputs.May 30May 30
Congratulations to Our 2025 GraduatesCDS introduces the class of 2025: a new generation of data scientists whose work spans ML theory, scientific computing, fairness, and more.May 16A response icon1May 16A response icon1
AI in Military Decision Support: Balancing Capabilities with RiskCDS Faculty Fellow Tim G. J. Rudner and colleagues at CSET outline responsible practices for deploying AI in military decision-making.May 14May 14
When Language Models Grade, the Average Score WinsAveraging LLMs’ full judgment distributions, instead of picking the most likely score, boosts grading accuracy without retraining.May 9May 9
When Good Data Is Scarce, Planning Beats Reinforcement Learning in AI Decision-MakingWhen AI can’t rely on good data, planning ahead beats traditional reinforcement learning.May 7May 7