Can cats see in the dark? Defending open knowledge through natural language search and information retrieval
🎥 Session recording: [https://youtu.be/u6tMzowZ10M?t=24365](https://youtu.be/u6tMzowZ10M?t=24365)
More and more, people are finding information through AI tools on search platforms and chatbots instead of visiting trustworthy sources directly. As more people encounter information through chatbots and algorithmic summaries, Wikipedia must evolve to remain a direct, trusted source for information retrieval. A part of this evolution includes working on the ability to make it easier and more intuitive to find things – whether by showing better answers to queries, or allowing people to ask queries in natural language rather than complex keywords. This session will explore how search on Wikipedia is adapting to this new reality. We will share research on internal search behavior, including how readers use descriptive queries and full questions, and what this reveals about shifting expectations. Participants will see a live demo of search experience allowing for natural language, or semantic, search and learn about early experimentation, key findings, and the trade-offs involved in balancing precision, discoverability, and trust.
In this hands-on workshop, attendees will test real-world use cases inspired by modern reader behavior: asking full questions, using natural language, and exploring descriptive search queries. Together, we will reflect on the results and brainstorm ways to improve information discoverability, navigation, and reliability while upholding Wikimedia’s core principles. The session invites collaborative discussion on how we defend open knowledge by strengthening Wikipedia’s role as a primary destination.
