Project Grant 2551618
- The National Science Foundation Division of Computing and Communication Foundations awarded the University of California, San Diego $600,000 on July 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop selective prediction techniques for visual-language models that can detect and avoid hallucinations. The project develops methods enabling large visual-language models to recognize uncertainty and abstain from making unreliable predictions in...
- The National Science Foundation Division of Computing and Communication Foundations awarded the University of Washington $600,000 on September 15, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop artificial intelligence methods that reduce hallucinations in large language models and improve their reliability, transparency, and uncertainty estimation for health information delivery. The project creates multilingual natural language processing...
- The National Science Foundation Division of Computing and Communication Foundations awarded Carnegie Mellon University $800,000 on October 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to study how users form and intensify beliefs about social understanding in conversational AI systems. The research examines how users attribute social understanding—perceived empathy, intention recognition, and genuine emotional comprehension—to chatbots and virtual...
- The National Science Foundation Division of Computing and Communication Foundations awarded the University of California, Berkeley $300,000 on September 15, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop artificial intelligence methods that reduce hallucinations in large language models and improve their reliability, transparency, and uncertainty estimation for health information delivery. The project develops multilingual natural language...
- The National Science Foundation Division of Computing and Communication Foundations awarded the University of Southern California $361,264 on June 15, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to conduct empirical and design research on human-AI interaction that preserves human critical thinking and creative agency. The project investigates how users engage in critical and creative thinking while collaborating with AI assistants on writing and...
- The National Science Foundation's Office of Multidisciplinary Activities awarded the University of Southern California $2,057,137 on July 1, 2026, under the Social, Behavioral, and Economic Sciences program (CFDA 47.075) to develop generative language models of neural activity driving conversational speech in longitudinal dialogues. The award funds the creation of a functional magnetic resonance imaging (fMRI) dataset capturing brain activity during interactive human conversations. Research...
- The National Science Foundation Division of Computing and Communication Foundations awarded the University of Illinois $999,999 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop novel techniques enabling large language models to succeed at complex mathematical and coding tasks while addressing inference efficiency concerns. The project focuses on diffusion-based LLMs that achieve global token awareness by modeling joint token...
- This Project Grant award from the National Science Foundation's Division of Information and Intelligent Systems (CFDA 47.070 - Computer and Information Science and Engineering) supports research into probing the inner workings of artificial intelligence (AI) systems and comparing them to human cognition during speech recognition. The $50,000 award to the University of Southern California (USC) will fund the development of novel mathematical models and experimental methods to examine how AI...
- The National Science Foundation (NSF) awarded a $800,000 Project Grant under the Computer and Information Science and Engineering (CFDA 47.070) program to the University of Illinois Urbana-Champaign. The 3-year grant, effective September 1, 2024, focuses on enhancing the safety of large language models (LLMs) used in high-stakes applications. The project aims to develop quantifiable safety measures and algorithms to detect and mitigate unsafe behaviors in LLMs, such as providing false or...
- This National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) Federal Grant Program (CFDA 47.070) award of $799,999 to the University of Washington provides funding for a 4-year collaborative research project focused on developing tools and methods to leverage large language models (LLMs) in support of divergent, convergent, and cooperative work. The key objectives of the project are to: 1) Develop design guidance and workflows for building more reliable and...
The National Science Foundation Division of Computing and Communication Foundations awarded the University of Southern California $450,000 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop techniques ensuring factual reliability in large language model chatbots. The project will create methods to prevent hallucinations in LLM-generated responses by leveraging interpretability of model internals. The work proceeds in three thrusts: the first will modify attention head contributions to ensure LLMs remain faithful to provided context documents; the second will analyze when model outputs are justified by internal knowledge through neuron localization; the third will integrate both approaches to handle questions requiring mixed contextual and knowledge-based reasoning. These techniques will enable real-time intervention to stop hallucinations as responses are generated, allowing LLM service providers to increase trustworthiness of their systems. Performance will occur in Los Angeles, California through July 31, 2029. The award is structured as a project grant.
Mod # | Description | ReasonForModification | Federal Obligation | Date |
|---|---|---|---|---|
| Not listed | $450.0k | 7/23/26 |