In this session, A+T 2020 speakers Martha Dark, Allison Gardner, and Kadija Ferryman discuss the U.K. Government’s roll-out technology as solutions to complicated societal problems and Foxglove’s work challenging the Covid-19 datastore; deploying AI responsibly in health tech; and tackling health disparities through data.
Martha Dark: Challenging the Covid-19 Datastore
Martha Dark, Foxglove, focuses on the U.K. Government’s roll-out technology as solutions to complicated societal problems and Foxglove’s work challenging the Covid-19 datastore. Martha explores the issues with some of these systems and what she thinks needs to change.
She also touches on Foxglove’s work challenging the U.K. Government’s A-level grading algorithm and the visa streaming tool.
Listen to our interview with Martha talking about making tech accountable. You can also read an edited version.
Allison Gardner: Responsible AI deployment: It isn’t just what you use but how you use it
Allison Gardner‘s talk takes discussion of ethical and responsible AI beyond discussion of the data and algorithm itself to the wider aspect of AI deployment:
- How we hard code in decision thresholds
- Design the interface
- Ensure meaningful human in the loop and feedback
- Consider operational procedures in implementing the system
- Undertake ongoing evaluation and audits
- Consider impact on workforce skills
Allison finishes by visiting the beginning of the development lifecycle by considering the questions we must ask ourselves when defining a problem and what we are aiming for by choosing an AI solution.
Listen to our interview with Allison talking about small nudges towards responsible AI. You can read an edited version of Allison’s conversation.
Kadija Ferryman: Health Data and Health Equity: Challenges And Opportunities
A cultural anthropologist and bioethicist, Kadija Ferryman‘s talk focuses on the recent US federal policy guidance for developing machine learning tools in medicine.
Kadija examines how this policy guidance fails to address the ways that health data reflect inequities, and she proposes modifications that can prevent group harms and advance more just health technology policy.
Listen to Kadija and Laura Sobola in conversation on The Human Show.