NYU Center for Data Science·4d agoA New Tool Gives Researchers Evidence to Back Their AI Training ClaimsA researcher can publish an AI model’s code and results and still leave readers unable to check how the model was actually trained. CDS…
NYU Center for Data Science·6d agoA Psycholinguistics Explanation of Why AI Writing Can Be Hard to ReadA sentence can follow the rules of English and still make readers struggle to connect its words. CDS Faculty Fellow Thomas Clark and his…
NYU Center for Data Science·Oct 28Testing AI for Failures Too Rare to Show Up in Safety ChecksA chatbot that gives a harmful answer once in 10,000 tries could pass hundreds of safety checks without a single failure. CDS Postdoc…A response icon1A response icon1
NYU Center for Data Science·Oct 23AskChem: A New Search Engine for Chemistry ResearchA chemist choosing how to make a molecule may need to compare reaction conditions and results scattered across dozens of papers. NYU…
NYU Center for Data Science·Oct 21AI Test Offers a Faster, Cheaper Way to Predict Breast Cancer RecurrenceSome of the most widely used tests for estimating whether breast cancer will return were developed some 25 years ago and require a lab to…
NYU Center for Data Science·Oct 16A World Model That Keeps Learning After Training EndsA robot can practice in one environment for millions of steps and still encounter something new when the lighting changes, an object weighs…A response icon1A response icon1
NYU Center for Data Science·Oct 9Meet the Faculty: Pratyusha SharmaThis entry is part of our Meet the Faculty blog series, which introduces faculty who have recently joined CDS.
NYU Center for Data Science·Oct 7Giving AI Models a Way to Adapt Their Priors In ContextA machine learning model can learn to make Bayesian predictions quickly, but assumptions learned during training can remain stuck inside…A response icon1A response icon1
NYU Center for Data Science·Oct 2CDS Researchers Challenge Anthropic’s Evidence That Language Models Can IntrospectIn 2025, Anthropic reported that its Claude models could detect when researchers injected a concept directly into their neural activity…A response icon1A response icon1
NYU Center for Data Science·Sep 30Meet the Faculty: Jaume Vives-i-BastidaThis entry is part of our Meet the Faculty blog series, which introduces and highlights faculty who have recently joined CDS.
