PETScan has long been one of Wikimedia’s most powerful discovery tools, helping contributors identify articles to create, improve, translate, and organize. During the June 2026 EduWiki Workshop, participants explored how its capabilities can be combined with generative AI to build more efficient editing workflows while maintaining the human judgment that underpins Wikimedia’s collaborative editing model.
Led by Mohammad Hijjawi, a Data Scientist, Full-Stack AI Engineer, active Wikimedian, and member of the Wikimedians of the Levant User Group, the workshop moved beyond theory through live demonstrations, practical editing scenarios, prompt engineering techniques, and discussions on responsible AI use across Wikimedia projects. 54 participants remained in the session for more than 30 minutes, representing Wikimedia communities and contributors from Sub-Saharan Africa (SSA), Northern and Western Europe (NWE), Central & Eastern Europe & Central Asia (CEECA), South Asia (SA), East, Southeast Asia and the Pacific (ESEAP), the Middle East and North Africa (MENA), and North America (NA). Live interpretation was available in Arabic, Chinese, French, Indonesian, and Spanish, enabling broader participation across language communities.
Finding the right place to contribute
Rather than beginning with artificial intelligence, the workshop started with the challenge many contributors face before making an edit: identifying where meaningful work is needed.
Through live demonstrations of PETScan, Mohammad introduced participants to different ways of discovering editing opportunities by searching categories, templates, Wikidata items, page properties, and maintenance categories. Participants learned how these searches can be used to create targeted worklists for article creation, translation, expansion, and quality improvement instead of relying on random article selection.
Several practical workflows were demonstrated, including identifying articles that exist in one language but are missing in another, locating short articles suitable for expansion, finding pages requiring citations, and discovering articles without Wikidata items. Throughout the demonstrations, emphasis was placed on creating repeatable workflows that could support edit-a-thons, education programmes, and newcomer activities.
Human-led AI workflows
Once editing opportunities had been identified, attention shifted to how generative AI could responsibly support different stages of the editing process. The workshop demonstrated how it could function as an assistant. Participants explored practical approaches for locating reliable sources, reviewing article structure, improving existing text, generating translation drafts, summarizing lengthy references, and even creating PETScan queries and SPARQL searches for Wikidata. Throughout each demonstration, the trainer reinforced the same principle:
Find with PETScan. Assist with AI. Verify as a human. Publish with confidence.
The session also examined the Wikimedia Foundation’s AI strategy together with English Wikipedia’s guidance on large language models, highlighting that contributors remain fully responsible for verifying every suggestion, checking every source, and ensuring that all published content complies with Wikimedia policies.
To support participants beyond the workshop, Mohammad also shared a collection of practical AI prompts developed specifically for Wikimedia editing. The prompts cover tasks such as reviewing article structure, evaluating citations, locating reliable sources, and improving article quality while maintaining Wikipedia’s standards for verifiability and reliability. These resources have been published alongside the workshop materials for continued learning.
Spotlight: Gap Atlas
One of the highlights of the workshop was the introduction of Gap Atlas, a new open tool developed by Mohammad Hijjawi that builds upon PETScan workflows to simplify content discovery across Wikimedia projects.
During the live demonstration, participants explored how Gap Atlas combines article statistics, including pageviews, references, article size, editor activity, and quality assessments, into a single interface, making it easier to prioritize articles for translation and improvement. The tool also allows contributors to export worklists for edit-a-thons and education programmes in multiple formats.
Gap Atlas is available at: https://gap-atlas.toolforge.org/
Participants responded positively to the demonstration and suggested several enhancements, including expanded multilingual support, additional quality filters, and improved workflows for smaller Wikimedia projects. Mohammad welcomed these suggestions and shared that the tool would continue evolving based on community feedback.
Community discussion and practical ideas
The workshop concluded with an active discussion that extended the live demonstrations into practical community use cases.
Participants explored how PETScan could help identify articles without images for Wikimedia Commons contributions, locate pages without Wikidata items in smaller Wikipedias, and support newcomer onboarding through targeted editing tasks. Suggestions were also shared on improving multilingual workflows and developing additional search capabilities for underrepresented language communities.
During the discussion, Mohammad committed to producing additional learning resources based on participant feedback, including a walkthrough for identifying articles without images, sharing the AI prompts demonstrated during the session, providing links to the Wikimedia Foundation’s AI strategy and English Wikipedia’s AI guidance, and continuing the development of Gap Atlas with community input.
Workshop resources
Resources from the workshop including Workshop presentation slides, Session recordings in all interpretation languages, and AI prompts shared by the trainer, are now available on our Workshop resources page on Meta-Wiki for anyone wishing to revisit the session or explore the tools further.
Continue the conversation at Wikimania 2026
If you are attending Wikimania 2026 this July, we would love to see you at two EduWiki Hub sessions.
Join us for our first in-person EduWiki Workshop, “Using Wikimedia in Education: A Practical Workshop by the EduWiki Hub,” taking place on Tuesday, 24 July, at 16:00 (Europe/Paris). The workshop will provide participants with practical strategies, tools, and examples for integrating Wikimedia projects into teaching and learning. Registration is available through the workshop’s Meta-Wiki event page, where you will also find the Eventyay link to add the session to your personal conference schedule.
You are also invited to this panel session, “AI and Wikipedia in Education: Opportunities, Ethics, and Equity in Practice,” taking place on 22 July 2026 at 11:00 (Europe/Paris). The panel will bring together Wikimedia and education practitioners to discuss the opportunities and challenges of integrating artificial intelligence into Wikipedia-based teaching and learning.
We look forward to continuing these conversations with the Wikimedia community at Wikimania 2026.
Get involved!
The EduWiki Hub invites you to take part in its growing initiatives:
Join the EduWiki community
Submit your Wikimedia education story to the Education Newsroom by the 26th of every month
Follow EduWiki Hub on social media Facebook, LinkedIn, and Instagram
Subscribe to the Education Newsletter to have This Month in Education delivered to you: Here
Explore the Hub’s monthly Open Educational Resources documentation
Collaborate with the Hub by contacting eduwikiug@gmail.com
Can you help us translate this article?
In order for this article to reach as many people as possible we would like your help. Can you translate this article to get the message out?
Start translation