' Workshop on participatory harm auditing of generative-AI for use in cultural heritage ' Workshop Data 2025 August

2421 15 8102 2 50114 document 8102 2421_ReadMe.txt text/plain fed54e9ab50f5add75c7ed92efe475df MD5 1011 2026-09-10 11:10:23 https://researchdata.gla.ac.uk/2421/1/2421_ReadMe.txt 2421 1 1 text/plain Text en public cc_by_4

2421_ReadMe.txt

readme 8103 1 50117 document 8103 indexcodes.txt text/plain af9dae6699351ae7547c3f5bfe7cc58c MD5 576 2026-09-10 11:10:34 https://researchdata.gla.ac.uk/2421/2/indexcodes.txt 2421 2 2 text/plain other Generate index codes conversion from Text to indexcodes en public

indexcodes.txt

http://eprints.org/relation/isVersionOf https://researchdata.gla.ac.uk/id/document/8102 http://eprints.org/relation/isVolatileVersionOf https://researchdata.gla.ac.uk/id/document/8102 http://eprints.org/relation/isIndexCodesVersionOf https://researchdata.gla.ac.uk/id/document/8102 8104 2 50118 document 8104 2421_Consent_Form.docx application/vnd.openxmlformats-officedocument.wordprocessingml.document d38eb3a467b5730968187b92e23c282d MD5 458437 2026-09-10 11:11:24 https://researchdata.gla.ac.uk/2421/3/2421_Consent_Form.docx 2421 3 3 application/vnd.openxmlformats-officedocument.wordprocessingml.document Text en public cc_by_4

2421_Consent_Form.docx

documentation 8105 2 50120 document 8105 2421_Participant_Information.docx application/vnd.openxmlformats-officedocument.wordprocessingml.document 2daa1e9c4b79c1291cbe99e22e598d42 MD5 485013 2026-09-10 11:11:28 https://researchdata.gla.ac.uk/2421/4/2421_Participant_Information.docx 2421 4 4 application/vnd.openxmlformats-officedocument.wordprocessingml.document Text en public cc_by_4

2421_Participant_Information.docx

documentation 8106 1 50124 document 8106 indexcodes.txt text/plain f9fe35652de1375044f28b1b828e9c9b MD5 510 2026-09-10 11:12:06 https://researchdata.gla.ac.uk/2421/5/indexcodes.txt 2421 5 5 text/plain other Generate index codes conversion from Text to indexcodes en public

indexcodes.txt

http://eprints.org/relation/isVersionOf https://researchdata.gla.ac.uk/id/document/8104 http://eprints.org/relation/isVolatileVersionOf https://researchdata.gla.ac.uk/id/document/8104 http://eprints.org/relation/isIndexCodesVersionOf https://researchdata.gla.ac.uk/id/document/8104 8107 1 50125 document 8107 indexcodes.txt text/plain f5982475386b1a8e057bc2fc57879449 MD5 2664 2026-09-10 11:12:07 https://researchdata.gla.ac.uk/2421/6/indexcodes.txt 2421 6 6 text/plain other Generate index codes conversion from Text to indexcodes en public

indexcodes.txt

http://eprints.org/relation/isVersionOf https://researchdata.gla.ac.uk/id/document/8105 http://eprints.org/relation/isVolatileVersionOf https://researchdata.gla.ac.uk/id/document/8105 http://eprints.org/relation/isIndexCodesVersionOf https://researchdata.gla.ac.uk/id/document/8105 archive 32963 disk0/00/00/24/21 2026-09-21 15:21:38 2026-09-21 15:21:50 2026-09-21 15:21:38 data_collection show Simpson Kate 55222 0000-0002-6883-4146 Soufan Ayah O'Hara Emily 78055 Bartliff Zoe 38396 0000-0002-1644-3151 Kim Yunhyong 12984 0000-0001-5400-0389 Ruthven Ian Moshfeghi Yashar 13082 glaresearchdata:2026-09-10-2421 ' Workshop on participatory harm auditing of generative-AI for use in cultural heritage ' Workshop Data 2025 August 10303000 30200000 Generative AI, auditing, participatory, cultural heritage. This workshop explored the ways in which CH professionals might audit the use of generative AI tools when used for the enrichment of descriptive metadata. For this workshop, we focused on the use of two AI tools/models as talking points to contextualise stakeholders’ thinking regarding generative AI and associated harms/audit (e.g. Meta’s Llama 2 and Google’s Gemini 1.5). These tools identify and/or describe the contents of images in a way that is concomitant with user facing CH metadata (archive catalogues, museum labels etc.). There are innumerable potential tools and uses for Generative AI within the CH workflow, however, the generated outputs gained from this workshop allow for deeper discussion across the CH sector regarding potential auditing methods for (and to a lesser extent, the use of) AI within CH workflows and policies. The data records participants’ background and their understanding of generative AI as well as how to recognise (and discuss the auditing of) potential risks or negative biases within both AI models and CH collections. The data is organised into survey collected on participant backgrounds and their perception of Generative AI, followed by online Miro board screenshots organised into workshop group discussion activities. 2026-09-10 University of Glasgow 10.5525/gla.researchdata.2421 Presentation Spreadsheet Text University of Glasgow and University of Strathclyde 324272 Participatory Harm Auditing Workbenches and Methodologies Simone Stumpf UK Research and Innovation ( UKRI) (UKRI) EP/Y009800/1 Computing Science TRUE English 2025-08-27 2025-08-27 2036-09-10 D Data will have to be destroyed after 10 years as per PIS. 2026-09-10 AA: "Aggregated data will be archived on the University of Glasgow’s institutional open-access research repository (Enlighten) for a period of 10 years". file_share TRUE 2730 other Approving committee:  University of Strathclyde, Computer and Information Sciences FALSE