Crowdsourced AI Audits and Understanding AI Harms

While teams of experts audit AI tools in the lab, already-deployed AI systems have risks that are difficult to measure a priori. Working directly with communities to crowdsource data about AI impact is crucial to ensuring future systems are transparent and accountable. 

To understand the full risks of deploying AI systems, we also need new, nuanced methods of investigating bias and harms, such as in how models make normative judgments

Data Tools for Workers

In a working reality that is increasingly algorithmically managed, workers, researchers, and advocates need tools to manage data and develop alternate algorithmic futures.

• Workers Algorithm Observatory 
• WeClock
• Drivers Seat

Data Rights as Labor Rights

While workplaces are increasingly quantified and surveilled, normative and legal notions of data privacy conflict with property, trade secret, and IP law in ways that limit worker data access. We need new analyses and understandings of how international data privacy and AI law should be applied in the labor context. 

Commercial Surveillance

The increasing ubiquity of corporate surveillance networks like Amazon's Ring Neighbors network carries real risks to public safety, privacy, and corporate power. Mapping these networks and understanding their impact using data and online experiments is one way to investigate those risks.