This live panel discussion is part of the AI Happy Hour series, brought to you by Stanford AIMI and friends. We cover hot topics in AI in medicine as well as live questions & comments from attendees. It's casual, insightful, and open to all!

Panelists:
Nestor Maslej - Stanford
Daniel Zhang - Stanford
Akshay Chaudhari - Stanford
Alaa Youssef - Stanford

Outline:
00:00 - Intros

02:53 - Generating the AI Index - fifth annual report, collaboration with academic, private and nonprofit entities, government, policy, identify new metrics such as diversity, technical AI ethics, internal novel analysis to supplement industry analysis

07:59 - Thoughts on ethics and AI - translating academic ethics work to practical use in health equity and precision medicine

10:38 - Policy and legislation - intent vs practical ethics, incentive alignment, EU AI Act, 5 year increase across all 25 surveyed countries, investment vs risk management, stratifying risk, system by system vs sector by sector approach, FDA/FAA, proactive vs reactive, E.G. GDPR, US vs EU

18:08 - FDA, Safety, Ethics - More detailed documentation, systematized model updates, transparency

21:04 - Large language models - Soccer fields of computers, training cost millions/billions of dollars, government/industry/academia collaboration and aligning incentives, English dominant NLP, benchmarks: lab vs real world, lack of diversity in ImageNet, AI Colonialism, under diagnose underserved populations, performance vs ethics benchmarks, political adversarial considerations

33:18 - Improving known shortcomings - algorithm vs data bias solutions, distributive justice, adaptive models for populations, institutional shift, computer scientists vs medical doctor user

38:47 - AI Investments - private investment doubled from 2020, number of companies decreasing, more difficult for startups to break in across all sectors, healthcare tops investment since 2017.

42:05 - Post market safety evaluation - role/entity need, metrics: over time/population, inter institution comparison, model output vs outcome, institutional available resources

48:08 Surprises from AI Index - industrial contributions to ethics, China and US collaboration on papers

51:03 - New ideas to investigate - popular opinion on trust, public investment, auditing, AI malpractice, diversity metric on training, accountability governance

Stanford AIMI: aimi.stanford.edu/
Twitter: twitter.com/StanfordAIMI

#StanfordAIMI #AI