This task focuses on leveraging our research skills to help the Wikimedia community (most directly WMF but also the broader editor community) adapt to the new AI capabilities emerging via a few complementary streams of work:
- Community engagement: beginning more engagement with community members to better understand potential use-cases and pain points / concerns with using AI on the Wikimedia projects. The core focus of this will be around the Wikimedia Hackathon in May.
- Guidance on 3rd-party ML: I'm helping to chart out a path for how WMF should work with 3rd-party ML services and working on providing guidance that can help product teams in making good decisions around these services.
- Piloting new models: most relevant is my work supporting the Android team in deploying a generative AI model for Wikidata descriptions but this line generally will consider what other prototypes can help us better understand how to appropriately incorporate these more powerful models into Wikimedian workflows.
Note that this goal originally was going to have a component that focused on data gaps but that work will now hopefully occur as part of my edit types project. This shift has been made in response to the emergent need to address questions / interest around generative AI over the past few months and shifts in annual planning.