Wellll boy that escalated quickly.
Ironically, the subject of my lightning talk at this event was that long-time wikipedians feel intimidated to talk about how they think we can use LLMs to advance the values and goals of the movement, which is bad for the movement because we're fighting bad-faith AI with one hand tied behind our backs.
The in-person response was positive and constructive, including many new demos of useful LLM-based or LLM-built tools. It also included "literally devoted their entire adult lives to Wikipedia" folks saying "yup, I built useful LLM tool X at a hackathon but haven't talked about it online because I know I'll get attacked".
Thanks for proving my point, I guess.
Talked to Wikipedians about AI all day. A thing that came through loud and clear: editors are experiencing AI as an outright high-volume assault. That is related to, but much more intense than, “I tried AI in 2024 and it sucks”. People building AI tools for wikipedians have to grapple with that.