Yolanda Gil

President (2024-2026) — Association for Computing Machinery Principal Scientist, Senior Director for AI and Data Science Initiatives — USC Information Sciences Institute Director, Center on Artificial Intelligence for Health — USC Information Sciences Institute Member (2024-2030) — National Science Board Past President (2018-2020) — Association for the Advancement of Artificial Intelligence

Listen — profile

0:00 / 2:25

Profile

Yolanda Gil has spent her career making an argument that’s suddenly fashionable but which she was making decades before the current AI boom: that the biggest payoff from AI won’t come from chatbots or code assistants, it’ll come from accelerating scientific discovery. Spanish-born, trained at the Technical University of Madrid and then at Carnegie Mellon under Jaime Carbonell, she has been at USC’s Information Sciences Institute since 1992 — a rare case of someone picking an unfashionable research direction and just patiently outlasting the hype cycles.

At USC/ISI she leads the Knowledge Capture and Discovery group and directs the Center on AI for Health. Her technical work sits at an underexplored intersection: semantic workflows, provenance tracking, knowledge graphs, and automated hypothesis generation. Systems like DISK take research questions, map them to data and computational pipelines, and propose what’s worth reporting back to the scientist. It’s the unglamorous plumbing of AI for science — the stuff that actually has to work before a language model “reading a paper” means anything.

She was president of the AAAI from 2018 to 2020, chair of ACM SIGAI for two terms before that, and is a Fellow of ACM, AAAI, IEEE, AAAS, and the Cognitive Science Society — an unusually complete sweep. In 2024 she was appointed to the National Science Board, giving her a direct line into U.S. science policy. She ran for ACM president in 2024 and lost to Yannis Ioannidis, but her footprint across AI’s institutional machinery is hard to overstate.

For developers learning AI, Gil is worth following for a specific reason: she represents a different bet than the frontier-lab consensus. She thinks the important question isn’t whether models will replace humans, but whether we can build systems thoughtful enough to be real research partners — tools that know what they know, track where their conclusions came from, and don’t hallucinate their way through a lab notebook. If you care about AI doing something real in biology, geoscience, or medicine, her work is one of the more honest roadmaps.