PEPPERDINE TALK: SLIDES, PAPERS, AND LINKS Nick Vincent | Simon Fraser University | September 2026 Updated September 26, 2026 Towards a 'Fully Attested' AI Supply Chain: Developments in AI Evaluation and Implications for a More Human-Centered Data Ecosystem I'll probably take these materials down after a few days, but feel free to reach out to me any time: nvincent@sfu.ca START HERE ========== PowerPoint slides (view only; September 26 snapshot): https://1drv.ms/p/c/c94663507a43064d/IQAW-RSeGByjR7RYVNHwB7nvAU00Pe0EKxGo78fd5c9K5RE My website and publications: https://nickmvincent.com/ Guilded AI: https://guildedai.com/ Peasant Labs (coding-agent transcripts and data autonomy): https://peasantlabs.org/ Research group, People- and Data-centric Computing (PadComp): https://www.padcomp.org/ Collective Bargaining for Information: https://cb4i.org/ My Data Leverage blog: https://dataleverage.substack.com/ FEATURED READING ================ This paper on Jewish American authors and intellectual property: A Canary in the AI Coal Mine: American Jews May Be Disproportionately Harmed by Intellectual Property Dispossession in Large Language Model Training Heila Precel, Allison McDonald, Brent Hecht, and Nicholas Vincent. ACM CHI 2024. Examines the disproportionate use of Jewish American authors' work in LLM training datasets and the potential for unequal economic harms. https://arxiv.org/abs/2403.13073 https://doi.org/10.1145/3613904.3642749 NeurIPS paper on collective bargaining: Collective Bargaining in the Information Economy Can Address AI-Driven Power Concentration Nicholas Vincent, Matthew Prewitt, and Hanlin Li. NeurIPS Position Papers 2025. The case for data creators bargaining collectively over how AI systems use their contributions and how the resulting value is shared. https://arxiv.org/abs/2506.10272 https://cb4i.org/ Recent Data Leverage posts on evaluation: - The AI "Evaluation Crisis" Is an Opportunity to Get Data Flow Right Nick Vincent, May 7, 2026. Why evaluation creates an opening to rebuild relationships with data creators and improve attribution https://dataleverage.substack.com/p/the-ai-evaluation-crisis-is-an-opportunity - Attestation across the AI Supply Chain Nick Vincent, April 11, 2026. https://dataleverage.substack.com/p/attestation-across-the-ai-supply PAPERS SHOWN DIRECTLY IN THE SLIDES ================================== These are the 11 papers pictured by title in the 35-slide snapshot. The next section supplies reading for indirect references and background topics, with the connection to the talk stated for each entry. 1. If open source is to win, it must go public Joshua Tan, Nicholas Vincent, Katherine Elkins, Magnus Sahlgren, Joseph Low, David Pham, Sampo Pyysalo, and Jenia Jitsev. ICML 2026. Slide 4. https://arxiv.org/abs/2507.09296 2. Overreliance in Writing Tasks: Exploring Similarity-Based Measures of AI Influence on Writing and Proposing a Reflective Writing Interface Intervention Vitor H. A. Welzel and Nicholas Vincent. ACM FAccT 2026. Slide 4; also relevant to slide 31. https://arxiv.org/abs/2605.15322 3. Mechanism Plausibility in Generative Agent-Based Modeling Patrick Zhao, David Huu Pham, and Nicholas Vincent. ACM FAccT 2026. Slide 4. https://arxiv.org/abs/2605.12824 4. An Audit and Analysis of LLM-Assisted Health Misinformation Jailbreaks Against LLMs Ayana Hussain, Patrick Zhao, and Nicholas Vincent. AIES 2025. Slide 4. https://arxiv.org/abs/2508.10010 5. Responsible AI in the OSS: Reconciling Innovation with Risk Assessment and Disclosure Mahasweta Chakraborti, Bert Joseph Prestoza, Nicholas Vincent, Vladimir Filkov, and Seth Frey. AIES 2025. Slide 4. https://ojs.aaai.org/index.php/AIES/article/view/36567 Earlier preprint, titled "Responsible AI in Open Ecosystems": https://arxiv.org/abs/2409.19104 6. Algorithmic Collective Action with Two Collectives Aditya Karan, Nicholas Vincent, Karrie Karahalios, and Hari Sundaram. ACM FAccT 2025. Slide 4. https://arxiv.org/abs/2505.00195 7. How Creatives Approach GenAI Image Generation: Tensions Between Structured Guidance, Self-Experimentation, and Creative Autonomy Haidan Liu, Isabelle Kwan, Taiga Okuma, Jeffrey Loverock, Nicholas Vincent, and Parmit K. Chilana. ACM Creativity & Cognition 2026. Slide 4. https://arxiv.org/abs/2605.10898 8. Tracing Everyday AI Literacy Discussions at Scale: How Online Creative Communities Make Sense of Generative AI Haidan Liu, Poorvi Bhatia, Nicholas Vincent, and Parmit Chilana. ACM CHI 2026. Slide 4. https://arxiv.org/abs/2603.09055 9. Examining Wikipedia With a Broader Lens: Quantifying the Value of Wikipedia's Relationships with Other Large-Scale Online Communities Nicholas Vincent, Isaac Johnson, and Brent Hecht. ACM CHI 2018. Slide 7. https://doi.org/10.1145/3173574.3174140 10. "Data Strikes": Evaluating the Effectiveness of a New Form of Collective Action Against Technology Companies Nicholas Vincent, Brent Hecht, and Shilad Sen. The Web Conference 2019. Slide 7; also relevant to slides 6 and 33. https://doi.org/10.1145/3308558.3313742 Open PDF: https://www.nickmvincent.com/static/www2019_datastrike.pdf 11. Collective Bargaining in the Information Economy Can Address AI-Driven Power Concentration Nicholas Vincent, Matthew Prewitt, and Hanlin Li. NeurIPS Position Papers 2025. Slide 7; project linked on slide 18. https://arxiv.org/abs/2506.10272 https://cb4i.org/ INDIRECT REFERENCES AND BACKGROUND READING ========================================= These papers connect to concepts, projects, and figures discussed in the talk. Connections identified from broad topic mentions are labeled "Background" rather than presented as explicit slide citations. 12. Data Leverage: A Framework for Empowering the Public in its Relationship with Technology Companies Nicholas Vincent, Hanlin Li, Nicole Tilly, Stevie Chancellor, and Brent Hecht. ACM FAccT 2021. Background for the data-strike, poisoning, and contribution framework on slide 6 and the collective-action discussion on slides 7-8. https://arxiv.org/abs/2012.09995 13. Can "Conscious Data Contribution" Help Users to Exert "Data Leverage" Against Technology Companies? Nicholas Vincent and Brent Hecht. ACM CSCW 2021. Background for redirecting contributions to preferred organizations and simulating recommender-data counterfactuals (slides 6-7). https://doi.org/10.1145/3449177 Open PDF: https://www.nickmvincent.com/static/cdc_cscw.pdf 14. Measuring the Importance of User-Generated Content to Search Engines Nicholas Vincent, Isaac Johnson, Patrick Sheehan, and Brent Hecht. AAAI ICWSM 2019. Background for observational measurement of human-created content's value to platforms (slide 7). https://doi.org/10.1609/icwsm.v13i01.3248 15. A Deeper Investigation of the Importance of Wikipedia Links to Search Engine Results Nicholas Vincent and Brent Hecht. ACM CSCW 2021. Background for the Wikipedia and search-audit research on slide 7. https://doi.org/10.1145/3449078 Open PDF: https://www.nickmvincent.com/static/wikiserp_cscw.pdf 16. The Dimensions of Data Labor: A Road Map for Researchers, Activists, and Policymakers to Empower Data Producers Hanlin Li, Nicholas Vincent, Stevie Chancellor, and Brent Hecht. ACM FAccT 2023. Background for data labor and knowledge-worker bargaining (slides 5 and 14-18). https://arxiv.org/abs/2305.13238 17. Pika: Empowering Non-Programmers to Author Executable Governance Policies in Online Communities Leijie Wang, Nicholas Vincent, Julija Rukanskaite, and Amy X. Zhang. ACM CHI 2024. Background for tools that help communities write governance policies (slide 3). https://arxiv.org/abs/2310.04329 18. IDP-Bench: Benchmarking ability of LLMs to protect personal information in interdependent privacy contexts Ayana Hussain, Soumya Sharma, Golnoosh Farnadi, Nicholas Vincent, Heber Hwang Arcolezi, and Ulrich Aivodji. Preprint, 2026. Background for the student-led benchmark work mentioned on slide 4; the slide does not name this benchmark. https://arxiv.org/abs/2606.09908 19. The Economics of Maps Abhishek Nagaraj and Scott Stern. Journal of Economic Perspectives 34(1), 196-221, 2020. Slide 15 says "Cf. economics of maps" and pictures the cartography essay below, which explicitly draws on this paper. https://www.aeaweb.org/articles?id=10.1257/jep.34.1.196 20. Measuring AI Ability to Complete Long Software Tasks Thomas Kwa et al. NeurIPS 2025; updated preprint, 2026. Research behind METR's task-completion time-horizon approach. Slide 19 pictures METR's time-horizon chart; the live chart has continued to change since the original paper. https://arxiv.org/abs/2503.14499 https://metr.org/ 21. Understanding Black-box Predictions via Influence Functions Pang Wei Koh and Percy Liang. ICML 2017. Foundational background for the influence functions mentioned on slide 33; the slide does not cite a particular paper. https://arxiv.org/abs/1703.04730 22. HealthBench: Evaluating Large Language Models Towards Improved Human Health Rahul K. Arora et al. Preprint, 2025. Background for the physician-led evaluation example: slide 21 mentions ChatGPT Health, whose announcement links to HealthBench. https://arxiv.org/abs/2505.08775 Ongoing work: games with a purpose for mechanistic interpretability is mentioned on slide 4. No public paper yet. ESSAYS, REPORTS, AND OTHER RESOURCES FROM THE TALK ================================================ AI Technologies are System Maps, and You are a Cartographer Nick Vincent, February 3, 2023. Pictured on slide 15. https://dataleverage.substack.com/p/ai-technologies-are-system-maps-and-you-are-a-cartographer Matthew Prewitt's writing (acknowledged on slide 18): https://mattprewitt.substack.com/ OpenMined: Attribution-Based Control (slide 24): https://openmined.org/attribution-based-control/ OpenMined's publisher discussion, also linked on slide 24: https://openmined.org/for-publishers/ GPT-5.5 System Card (OpenAI, April 2026; pictured on slide 19): https://openai.com/index/gpt-5-5-system-card/ Evaluation organizations mentioned on slides 19, 21, and 29: METR: https://metr.org/ Apollo Research: https://www.apolloresearch.ai/ UK AI Security Institute (AISI): https://www.aisi.gov.uk/ Introducing ChatGPT Health (OpenAI, January 2026; slide 21): https://openai.com/index/introducing-chatgpt-health/ HealthBench overview: https://openai.com/index/healthbench/ Artificial Analysis coding-agent benchmarks (slide 14): https://artificialanalysis.ai/agents/coding-agents News Corp and OpenAI partnership announcement (May 2024; slide 23): https://openai.com/index/news-corp-and-openai-sign-landmark-multi-year-global-partnership/ Benchmarks 101 David Huu Pham, Heila Precel, Eleanor Tursman, B Cavello, Francisco Jure, and Nicholas Vincent. NeurIPS 2025 Education Program. Additional educational reading for the benchmarking discussion. https://openreview.net/forum?id=TSkYNaS4RA SELECTED NEWS AND ANNOUNCEMENTS DISCUSSED ======================================== These links accompany the news examples in the deck. Dates refer to the original coverage or announcements, not to the date you read this file. OpenAI: Priorities and principles for effective third party assessments September 22, 2026. Primary announcement behind the slide 9 coverage. https://openai.com/index/priorities-principles-third-party-assessments/ Anthropic: Partnering with Accenture on embedded evaluation September 18, 2026. Pictured on slide 9. https://www.anthropic.com/news/accenture-embedded-evaluation AI needs its own accident investigators Chris Stokel-Walker, Fast Company, September 22, 2026. Slide 9. https://www.fastcompany.com/91609706/ai-needs-its-own-accident-investigators OpenAI to let third parties evaluate AI models during training Quartz, September 23, 2026. Headline pictured on slide 9. https://qz.com/openai-third-party-safety-evaluations-training-092326 Additional coverage: The New York Times v. OpenAI and Microsoft Microsoft exec called AI scraping the "largest theft of labor in human history" Ashley Belanger, Ars Technica, September 17, 2026. Reports on internal messages cited in the newly unsealed summary-judgment motion by news plaintiffs, including Brent Hecht's quoted warning. https://arstechnica.com/tech-policy/2026/09/microsoft-exec-called-ai-scraping-the-largest-theft-of-labor-in-human-history/ Allegations of AI distillation spark debate about IP theft, but is it illegal? NPR, July 28, 2026. URL supplied on slide 16. https://www.npr.org/2026/07/28/nx-s1-5909652/allegations-of-ai-distillation-spark-debate-about-ip-theft-but-is-it-illegal For more publications and future updates: https://nickmvincent.com/