Projects

Lightweight, easy-to-train adapters that allow communities to customize LLMs can help decentralize AI use and help communities experiment with how to best use AI systems.

HS

Through ethnographic and qualitative work, we are working to understand how and when marginalized groups like the LGBTQ+ community successfully develop community-driven platforms that are more aligned with their values and needs.

HS

We are studying how online platforms' algorithms and policies shape the material conditions and labor of content creators and cultural workers.

BR

We are building new crowdsourcing tools that will allow users of platforms, ranging from Instagram to TikTok to Doordash, donate data to researchers and advocates with minimal effort.

VR

AG

SD

Police departments across the US are increasingly using AI tools for surveillance and automation of police work, such as filing police reports from body cam footage. We are working to evaluate the risks and failures of using AI in high-risk scenarios like law enforcement.

MN

NE

We are using a corpus of hundreds of articles and press releases about AI use in law enforcement automation and surveillance as a case example to theorize and empirically examine how AI 'hype' spreads

Investigating how users internally negotiate the use of AI in their daily lives to design more contextually appropriate systems.

Current approaches to understanding AI behavior are limited to short or few-shot interactions. We are building platforms and doing field work to understand how model behaviors like sycophancy and mimesis change in realistic, long-term interactions.

The Workers Algorithm Observatory (WAO) is a inter-institutional collaboration focused on developing tools and doing research to support labor organizing and advocacy in the algorithmic age.

SD

VR