Representatives from academia, industry and government gathered September 8-10 for a University of Florida workshop focused on prescriptive artificial intelligence for decision making.
A common application of AI research is predicting future life events, such as the risk of developing a particular health condition or the likelihood of making a job transition, based on current and past information. The concept of prescriptive AI, however, goes beyond predictions to suggest strategies that may influence events. This may include the use of prescriptive AI for drug repurposing as well as behavioral or policy changes.
Now in its fourth year, the UF Data & Artificial Intelligence Symposium, or DAISY, workshop attracted participants from around the world to discuss prescriptive AI’s potential and ethical use.
“We found a common thread across all areas we explored, including medicine, public health, pharmacy, economics and psychology: prescriptive AI is tied to changing the future for the better, which means making the best guesses on unseen worlds — it is imaginative!” said workshop co-organizer Mattia Prosperi, Ph.D., the associate dean for artificial intelligence and innovation at the UF College of Public Health and Health Professions and a professor in the department of epidemiology. “However, prescriptive AI also bears the responsibility of the consequences of automated decisions, as they could produce unintended harm in the same way predictive AI can be discriminatory.”
This year’s DAISY workshop featured five keynote speakers and multiple panels and theme talks. Large language models, or LLMs, (the most well-known of these is ChatGPT) featured prominently in several talks, including by Danielle Belgrave, Ph.D., the vice president of AI and machine learning at pharmaceutical company GSK. She described how GSK has fine tuned their own large language model to identify compounds from medical literature or other sources. Andy Lin, vice president of strategy and chief technology officer for Mark III Systems – NVIDIA, showcased how the next-generation chipsets will allow even more powerful LLMs. Moontae Lee, Ph.D., an assistant professor of information and decision sciences at the University of Illinois Chicago, provided insights on how soon LLMs might become capable of reasoning logically beyond their current limitations. Tianjun Sun, Ph.D., an assistant professor of industrial-organizational psychology and quantitative methods at Rice University, showed cutting edge applications of LLMs in psychometrics that are able to capture in new, unprecedented ways the complexity of personality and behavioral traits.
Presenters such as Jason Moore, Ph.D., chair of computational biomedicine at Cedars-Sinai Medical Center, and Patrick Ryan, Ph.D., vice president of observational health data analytics at Janssen, discussed the automation of AI processes for research studies, including their standardization to ensure replicability.
“What stood out to me is that while diverse content was presented and we had different expertise and backgrounds, all the presenters and many attendees I chatted with value mostly the same things: rigor, fairness and utility,” said Aprinda Indahlastari Queen, Ph.D., an assistant professor in the PHHP department of clinical and health psychology and a DAISY workshop presenter. “We all strive to continue improving current AI methods to be more accurate, applicable and shareable to the public.”
Looking ahead to DAISY 2025, Prosperi expects more focus on causality of AI for decision making, based on the many discussions on this topic at this year’s workshop, as well as more corporate involvement and sponsorship.
“We had a strong international representation, and some rumors say it could be organized abroad next year!” Prosperi said.
DAISY 2024 sponsors included:
- UF College of Public Health and Health Professions
- Department of health outcomes and biomedical informatics, UF College of Medicine
- Department of pharmaceutical outcomes and policy, UF College of Pharmacy
- UF Warrington College of Business
- Center for Cognitive Aging and Memory, UF McKnight Brain Institute
- Office of Research Affairs, UF College of Medicine – Jacksonville
- Endowed Chair, UF Emerging Pathogens Institute; UF department of pathology, immunology, and laboratory medicine
- UF Office of Research
- NVIDIA/Mark III Systems