Overview

As AI continues to dominate public attention and private investment, there is more regulatory scrutiny than ever on every point in the “AI stack”: from supply chain and computing needs, to model infrastructure, to downstream applications. Of particular interest to policymakers is the data that defines many AI products, and operates as a key differentiating factor from other computing technologies. In the modeling pipeline, data plays a key role in determining the worldview of the models, defining model performance claims and tasks in evaluations, shaping standards development, capturing human preferences in feedback, and communicating key details about the system.

Data, in particular, throws up a range of policy challenges. From competition (Is data becoming an unavoidable barrier to entry?); privacy (What are the unique privacy risks posed by LLMs? Are they solvable?); labor (What are the working conditions for the human workers training and improving AI systems?); transparency (Should access to training data be a regulatory prerequisite to ensure accountability?); copyright, and more. It also raises more fundamental questions like the sustainability of the “bigger is better” paradigm in AI, especially when such a paradigm incentivizes the often invasive collection of data about people and communities (Bender et al. 2021, Buolamwini 2023, Crawford 2021, Kak and Myers West 2023, Raji et al. 2020).

The Social Science Research Council (SSRC) invites applications for its research development workshop on Drilling Down to the Data: Navigating Data Politics at the Heart of AI Policy on July 22–23, 2024. 

The workshop will synthesize perspectives across different knowledge communities, including university researchers, legal scholars, policymakers, community organizers, technologists, and artists, to support grounded, policy-relevant research.

Applications

We particularly encourage applications from early-career scholars. We welcome applications from across various fields and practices, including the arts, the social sciences, the humanities, legal studies, journalism, community organizing, and data and computer science.

Proposals will be evaluated based on (1) relevance to the workshop’s theme, (2) whether they are at a stage of development where feedback from peers will make the most impact, and (3) complementarity with other research topics selected for the workshop.

Three weeks before the workshop, participants will share in-development manuscripts with all participants and our two chairs responsible for guiding the discussion. Participants are expected to submit a 5-page reflection and read each other’s work before the workshop to use the in-person time for discussion and feedback.