Integrating Machine-Generated Short Descriptions into the Wikipedia Android App: A Pilot Deployment of Descartes
This addresses content gaps for Wikipedia editors and users across multiple languages, though it is an incremental deployment of an existing model.
The researchers tackled the problem of uneven coverage of short descriptions in Wikipedia by piloting Descartes, a multilingual model, in the Android app, where 90% of accepted machine-generated descriptions were rated at least 3 out of 5 in quality, comparable to human-written ones.
Short descriptions are a key part of the Wikipedia user experience, but their coverage remains uneven across languages and topics. In previous work, we introduced Descartes, a multilingual model for generating short descriptions. In this report, we present the results of a pilot deployment of Descartes in the Wikipedia Android app, where editors were offered suggestions based on outputs from Descartes while editing short descriptions. The experiment spanned 12 languages, with over 3,900 articles and 375 editors participating. Overall, 90% of accepted Descartes descriptions were rated at least 3 out of 5 in quality, and their average ratings were comparable to human-written ones. Editors adopted machine suggestions both directly and with modifications, while the rate of reverts and reports remained low. The pilot also revealed practical considerations for deployment, including latency, language-specific gaps, and the need for safeguards around sensitive topics. These results indicate that Descartes's short descriptions can support editors in reducing content gaps, provided that technical, design, and community guardrails are in place.