Sabin Faculty Affiliate Sarra Alqahtani on Making Agentic AI Safer
Exciting news and foreboding warnings abound alongside the accelerated development of Artificial Intelligence these days. Multi-Agent AI (MAAI), which consists of networks of AI agents operating together in a dynamic environment, is no exception. Sabin Center Faculty Affiliate Sarra Alqahtani recently received a National Science Foundation CAREER award to help advance the safety and reliability of MAAI through an approach known as Multi-Agent Reinforcement Learning (MARL). Alqahtani explains how her MARL agents bolster security through learning:
They think, they reason and they make decisions based on the dynamic environment around them. So they don’t only predict, they predict and they make decisions based on that prediction. They also identify the uncertainty level around them and then make a decision about that: Is it safe for me to make a decision or should I consult a human?
Training these MARL tools requires optimizing them for unexpected environments and sequences, a process that is typically undertaken in gaming simulations where real-world risks are avoided. But Alqahtani is working to advance their safe and predictable functioning for real-world scenarios to prevent harms to the spaces and places in which they operate, as well as the users who will rely on them.
Alqahtani tests her MARL algorithms on drones flying over the Peruvian Amazon working to detect illegal gold mining in the rainforests. Partnering with hardware engineers, researchers in biology, ecology, and other environmental sciences, including those on the front-lines of addressing illegal gold mining, Alqahtani improves the efficacy and safety of these critical drone operations.
Professor Alqahtani explains how her work helps to make the future of AI safer for us in this brief but informative video.
Learn more about Professor Alqahtani’s work in the Peruvian Amazon, including how she provides hands-on learning for students while partnering with practitioners in the field.
And to get a better sense of the role AI and related technologies play in advancing our vision of a more sustainable, just, and abundantly wild world, check out our work in AI for a Changing World.