UCLA Computer Scientist Receives $2.8M DARPA Grant to Demonstrate New AI Model

Guy Van den Broeck, a professor of computer science at the UCLA Samueli School of Engineering, has received a three-year, $2.8 million research grant from the Defense Advanced Research Projects Agency (DARPA) to lead the development of trustworthy artificial intelligence systems. Joining him as the study’s co-investigator is Kai-Wei Chang, an associate professor of computer science, who brings expertise on fairness and generative AI for natural language processing. 

The grant is part of the DARPA’s Assured Neuro Symbolic Learning and Reasoning program, which seeks to ensure AI systems will operate safely and perform tasks as assigned. The award will support Van den Broeck’s research into neuro-symbolic AI, an emerging field that combines two approaches used in AI: neural networks and symbolic AI. In neural networks, interconnected nodes function like connected neurons to mimic the brain, while symbolic AI processes information represented by symbols. This hybrid of neuro-symbolic AI combines the massive data-crunching, pattern-finding power of neural networks with symbolic AI’s ability to incorporate knowledge and context. 

Ultimately, findings from this research could offer a pathway toward AIs that can reason and make informed decisions based on analysis, while also reducing the energy-intensive consumption of powerful AI models that rely only on neural networks. The study also aims to reduce bias in AI and its use of harmful language. 

The project includes researchers from UC Irvine and the University of Texas at Dallas. The group will first build the underlying algorithms and architecture of neuro-symbolic AI, and then demonstrate its capabilities and effectiveness. Van den Broeck, Chang and their colleagues have previously succeeded with this approach in optimizing routes through video game world maps and solving mathematical puzzles. 

Van den Broeck, a Samueli Fellow at UCLA, heads the UCLA Statistical and Relational Artificial Intelligence Lab that researches machine learning, knowledge representation and reasoning. Chang leads the UCLA Natural Language Processing Group, which works to develop algorithms and models in generative AI.