Deepening Public Participation In Data And AI

Deepening Public Participation In Data And AI Presented by: Mr. Peppin Date: March 17, 2022 Introduction - There is a huge amount of focus on ensuring AI is 'ethical'. - But according to whose ethics? Whose values? Whose needs? - What will be discovered: • About the Ada Lovelace Institute • Public participation in data and Al • Two research projects:  Public dialogue on location data  Algorithmic Impact Assessments About Ada - An independent research institute with a mission to make data and Al work for people and society. - Established in 2018 by the Nuffield Foundation - Our namesake, Ada Lovelace, was a c19th mathematician. - She wrote an algorithm for Charles Babbage's Analytical Engine and is widely considered the first ever computer programmer. - For many, she also stands for the role of marginalized people in the past and present of computing - a woman whose contributions were integral but often overlooked. - Convening diverse voices - Building evidence - Shaping and informing good policy and practice Ada’s Research Programs - BiometricsS - Public-sector use of data and algorithms - The future of regulation - Health data and COVID-19 technologies - Ethics and accountability in practice Public engagement and participatory research at Ada - Public attitudes research - Deliberation & dialogue - Participatory ethnography - Participation in policy & practice Participatory Data Stewardship - The responsible use, collection and management of data in a participatory and rightspreserving way. - Trustworthy data use and effective governance - Fairer and more equitable outcomes - Broader social welfare outcomes supported by data - Increased social license and active participation - More representative, inclusive and proportionate uses Ada: - Inform • We will keep you informed about how your data is being governed • Consult • We will listen to, acknowledge and provide feedback on concerns and aspirations for the governance of your data - Involve • We will work with you to ensure your concerns and aspirations are directly reflected in data governance - Collaborate • We will look to you for advice and innovation in design of data-governance models, and incorporate your advice and recommendations where possible - Empower • We will advise and assist in line with your decisions about your own datagovernance models Inform - Transparent data use - Explain what data is being used, by who, why and how it's looked after. - Checkable, assessable: Anyone else can easily check what you've done. Consult - Beyond Face Value (2019) Ada Lovelace Institute. Involve - The Citizens' Biometrics Council (2020) Ada Lovelace Institute • Mini-public • 50 members of the public • 60+ hours of deliberation Collaborate - Camden Data Charter (2021) Empower - Open Humans Initiative - (some) Data Institutions Location Data Dialogue - In 2021 the Geospatial Commission commissioned a public dialogue on the ethical and trustworthy use of location data. It involved: • 85 people from across the UK • 4 online workshops – around 25 hours • Participant led research • 7 specialist presentations (ft. Jeremy!) Location Data Dialogue: Findings - Awareness and understanding - Data insecurity - Accountability, agency and governance - Reliance on and choices about data Location Data Dialogue: A Vision For Ethical And Trustworthy Location Data Use - Intent • How data is used and who benefits are key in considering ethics and trustworthiness - Accountability • - Transparency • - Data collectors should be accountable to regulators and data subjects Information on use, access, and storage should be succinct and accessible Agency • Consent should offer genuine choice to people about how their data is used Algorithmic Impact Assessments (AIAs) - Policymakers, Al developers and researchers are interested in algorithmic impact assessments, but there is little evidence on how they work in practice. - Working with NHS, we developed a process for assessing possible societal impacts of an algorithmic system before the system is in use. AIA process - A team seeks to use data for their Al system: 1. Reflexive exercise 2. Application filtering 3. Participatory workshop 4. Synthesis 5. Data access decision 6. Publication 7. Iteration AIA Participant Panel - We proposed a panel of patients and members of the public should be established. - This panel will review applications and their impact assessments, and identify other potential impacts. - This will inform data access decisions.