Using AI to generate synthetic data for SDGs monitoring
Tuesday, 26 March 2024
1:00 to 2:30 pm EST
UN Secretariat Building, S-1522 (open to all with a valid UN pass)
and online (Teams link will be shared with registered participants)
Register by 22 March at https://forms.office.com/e/MEpUL0uEhU
Artificial intelligence (AI) is increasingly being used to fill data gaps, in ways that hold a lot of promise for SDGs monitoring. As we approach the 2030 deadline, there are still many people – including those at greatest risk of being left behind – whose living conditions are not well captured by existing monitoring. These can include for example women, the elderly, and countries in the Global South, as well as people who are marginalised socioeconomically, live in hard-to-reach areas, or whose data has not been collected for other reasons. As new AI technologies are deployed globally, creating synthetic data of various kinds is becoming easier. However, these new technical opportunities come with certain risks – errors and biases which, if unaccounted for, can go against the objectives of SDGs monitoring and turn out to impede the design of effective policies.
In this Development Policy Seminar, Ms. Eleonore Fournier-Tombs, Head of Anticipatory Action and Innovation at UNU Centre for Policy Research, will cover the technical opportunities, as well as the pitfalls, of using AI to generate synthetic data, especially if this synthetic data then goes on to be used to train new AI models. The discussion, moderated by EAPD Director Shantanu Mukherjee, will draw from a recently published UNU policy report on the use of synthetic data to train AI models.
