Click on any of the links below to see the abstract provided by the instructors, located further down on this page.
All course times are each Pacific (UTC-7), each day from Tuesday July 28th through Thursday July 30th
E01: AI-assisted Literature Review: Building Open, Machine-Readable Literacy for Research and Education, Featuring the Semantic Climate Encyclopedia
Class Time 7:00 AM – 9:00 AM (Pacific Time – UTC-7)
Instructor Simon Worthington (TIB – Leibniz Information Centre for Science and Technology and University Library)
Affiliation #semanticClimate
Job Title Publishing Knowledge Graph Researcher
Course Abstract
The exponential growth of scholarly literature across disciplines has created major challenges for discovering, synthesizing, and reusing knowledge in a systematic and scalable manner. Traditional literature review practices and static reference resources are increasingly inadequate for handling large, heterogeneous, and rapidly evolving information landscapes. The course demonstrates how AI-assisted literature review workflows can be built, evaluated, and adapted using open access scholarly literature and transparent, reproducible methods.
The proposed framework leverages semantic technologies, knowledge graphs, and natural language processing to transform unstructured publications, reports, and documents into structured, interoperable, and reusable knowledge representations. AI-assisted literature review supports automated, transparent, and continuously updatable synthesis of evidence, enabling efficient navigation of large corpora and reducing manual effort. Alignment with shared vocabularies and standards ensures consistency, interoperability, and long-term reusability of the curated knowledge.
A core component of the framework is Named Entity Recognition (NER), which enables the automated identification and classification of key entities—such as species name, organizations, locations, diseases, drugs from research publications. NER provides the foundation for linking entities across documents, establishing semantic relationships, and populating knowledge graphs that support advanced querying, comparison, and reasoning.
Central to this effort is the Semantic Climate Encyclopedia—a dynamic, semantic reference that systematically enriches keywords with linked data from Wikimedia sources, and organizes them into a structured, searchable knowledge system. Semantic Encyclopedias can be automatically created for any corpus and are an excellent way for researchers to get an introduction to a new subject within an hour or two.
Together, the Semantic Encyclopedia, AI-assisted review, Named Entity Recognition (NER), and LLM-powered Retrieval-Augmented Generation (RAG) form a dynamic, extensible infrastructure that improves research efficiency, enhances knowledge accessibility, and supports data-driven learning and decision-making across disciplines. This integrated approach advances open science by bridging human-readable scholarship with machine-interpretable knowledge, enabling scalable, reusable, and interoperable research and educational ecosystems.
Course units include:
- Semantic corpus creation, text and data mining
- Named Entity Recognition (NER) for structured knowledge extraction
- Semantic encyclopedia for knowledge enrichment
- Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) for processing and querying scientific documents
The course will enable to build, deploy, and adapt an AI-assisted literature review workflow for semi-automated discovery, retrieval, and synthesis of peer-reviewed research articles. The instructional example uses scientific chapters and references from large open access article collections comprising millions of research articles. Participants will be introduced to core AI and machine learning concepts for literature analysis, including document retrieval, text mining, NER, and LLM-based retrieval systems, with a strong emphasis on evaluation, bias awareness, and the trustworthiness of AI-assisted outputs.
Contributors
Dr. Gitanjali Yadav, BRIC-National Institute of Plant Genome Research (BRIC-NIPGR) (Co-course chair); Prof. Peter Murray-Rust, Cambridge University (Co-course chair); Dr. Renu Kumari, #semanticClimate (Instructor), Ms. Priti, BRIC-National Institute of Plant Genome Research (Instructor), Ms. Udita Agarwal, BRIC-National Institute of Plant Genome Research (Instructor)
| Date | Title | Instructor | Duration |
| Day 1 | Knowledge extraction from scientific literature corpus | Dr. Renu Kumari, #semanticClimate, BRIC-NIPGR | 2 h |
| Day 2 | Create semantic encyclopedia for knowledge enrichment | Ms. Udita Agarwal, #semanticClimate, BRIC-NIPGR | 2 h |
| Day 3 | LLM and RAG for processing and querying scientific documents | Ms. Priti, Ms. Udita Agarwal, Dr. Renu Kumari #semanticClimate, BRIC-NIPGR | 2 h |
Audience: Researchers, librarians, publishers
Level: (Beginner, but suitable for all levels)
Requirements: Run Google Colab in a browser. See: https://colab.research.google.com/ A Google account to run CoLab Jupyter Notebooks
Course Learning Objectives
At the end of the course, participants will be able to:
- Conduct a scoping literature review using semantic toolkits.
- Be able to carry out text and data mining and build a corpus from open access sources such as EPMC
- Obtain familiarity with using LLM RAG with PDFs and with HTML
- Use Colab Jupyter Notebooks, execute python commands, and use GitHub.
- Use the provided ‘good practice’ framework for managing LLM projects
- Knowledge Graph
E02: Using the ORCID, OpenAlex, Open Policy Finder, and Unpaywall APIs in R to Harvest Institutional Data
Class Time: 7:00 AM – 10:00 AM (Pacific Time – UTC-7)
Instructor Danielle Kirsch
Affiliation Oklahoma State University
Job Title Research Data Services Librarian
Course Abstract
The objectives of this course are to (1) obtain a set of ORCID iDs for people affiliated with your institution, (2) harvest a list of DOIs from OpenAlex for publications associated with these iDs, and (3) gather open access information for the articles using Open Policy Finder (formerly Sherpa-Romeo) and Unpaywall. Students will work with a set of pre written scripts in R, customizing them for their institutions to access the APIs for ORCID, OpenAlex, Open Policy Finder, and Unpaywall, and bring it all together into a manageable data file. While some experience using R will be helpful, it is not required. However, much of the workshop time will be spent explaining fairly complex chunks of code, with periodic breakout sessions for attendees to modify relevant portions of their code, ask questions, and explore their results. Although the basics of using R will be reviewed briefly at the beginning, the emphasis of the course will be on running the scripts and gathering and interpreting the data. In other words, this course is not focused on learning R, but rather on obtaining a dataset of publications based on institutional affiliation and open access information on those publications. It is inspired by a course taught previously at FSCI, available at https://osf.io/vpgbt/. The course will conclude with a discussion on using this data to develop outreach methods to authors to inform them of their right to deposit author manuscripts.
Outline:
Day 1: Goals for the class, basic introduction to R, navigating the RStudio environment, installing libraries, troubleshooting, getting set up in Posit Cloud. Introduction to rorcid & the ORCID API, configuring rorcid, gathering ORCID IDs for my institution.
Day 2: ORCID continuation. Harvesting bibliographic metadata using openalexR, looking up open access information in Unpaywall, and exploring resulting data.
Day 3: Journal publisher policies with Open Policy Finder, tying up loose ends, putting it all together: what can we do with this data? Running summary information and basic plots, conducting outreach to faculty.
Contributors
Clarke Iakovakis, Scholarly Services Librarian, Oklahoma State University (contributor)
Target Audience
Researchers, librarians, faculty/scholars, publishers, administrators, technical support
Engagement
The course will be a mix of lecture (with the option for students to ask questions throughout) and interactive breakout sessions. During breakout sessions, students will customize portions of their code to retrieve results for their institutions and researchers, as well as have opportunities to receive troubleshooting support and ask questions. Each breakout session will be followed by open discussion. All coding scripts used during this course will be provided pre-written to attendees.
Main Topics
APIs, open access, ORCID, OpenAlex, Open Policy Finder (Sherpa Romeo), Unpaywall, research information management, persistent identifiers, repositories, scholarly publishing, R
Requirements
Students will not need to install anything for the course; we will be working from a cloud-based version of R Studio. Some programming/R experience would be helpful but
it is not required. Students will be required to have accounts with ORCID, Posit Cloud (formerly called R Studio Cloud), and Open Policy Finder (formerly Sherpa-Romeo), and to obtain API keys for ORCID and Open Policy Finder. Instructions for this will be provided to registrants.
E03: Who’s afraid of knowing? Critical thinking, epistemic responsibility, and open scholarship in the age of AI
Class Time: 7:00 AM – 10:00 AM (Pacific Time – UTC-7)
Instructor Francis P. Crawley
Affiliation GCPA & SIDCER
Job Title Executive Director
Course Abstract
Course description As part of the FORCE11 Scholarly Communication Institute in partnership with the UCLA Library (FSCI 2026), this course is designed to engage a global audience in a critical examination of how scholarly thinking, judgement, and responsibility must evolve as artificial intelligence increasingly mediates the full research and publication lifecycle. The course builds on the FSCI 2023, 2024, and 2025 courses addressing AI ethics, governance, authorship, and integrity in scientific publications, and advances the discussion by focusing explicitly on the human cognitive layer of open scholarship. Artificial intelligence is no longer confined to discrete tools for writing or data analysis. It now shapes discovery, synthesis, peer review, evaluation, ranking, and dissemination, subtly reconfiguring what is visible, legible, and valued in scholarly ecosystems. This three-day, three-hour-per-day course examines how critical thinking itself is transformed under these conditions, and how scholars can retain epistemic humility, judgement, and responsibility in AI-mediated open scholarship environments. Participants will engage in structured discussions and reflective exercises that
interrogate how epistemic authority, responsibility, and scholarly agency are redistributed when AI systems act as intermediaries in knowledge production and evaluation. Rather than focusing on technical mastery or compliance, the course foregrounds critical awareness, human judgement, and ethical reflexivity as core scholarly competencies in the age of AI. The human cognitive layer in AI-mediated scholarship A central focus of this course is the human cognitive layer: the ways in which scholars think, judge, interpret, and take responsibility when AI systems increasingly mediate scholarly work. While prior discussions on AI in research have emphasized ethics, governance, and policy, this course examines how AI reshapes epistemic practices themselves. Participants will explore how AI-mediated systems influence perception, attention, trust, and valuation in scholarly work, and how these influences affect originality, responsibility, and intellectual autonomy. The course emphasizes that while AI can automate tasks, it cannot replace epistemic judgement, contextual reasoning, or moral responsibility. Understanding where and how these human capacities remain essential is critical to sustaining trustworthy and meaningful open scholarship. In this course, research assessment systems, digital infrastructures, and AI enabled scholarly platforms are not treated as topics in their own right. Rather, they are approached as epistemic environments: conditions that shape how knowledge is encountered, evaluated, trusted, and acted upon. The focus of the course remains on how scholars think, judge, and take responsibility for knowing within these environments. Specific challenges for critical thinking in AI-mediated open scholarship The integration of AI into scholarly ecosystems presents a set of cognitive and epistemic challenges that extend beyond ethics and governance: 1. Delegation of judgement AI systems increasingly influence decisions about relevance, quality, and impact, raising questions about where human responsibility resides. 2. Epistemic opacity Probabilistic outputs and opaque models complicate scholars’ ability to evaluate claims, sources, and reasoning. 3. Normalization of automation Routine reliance on AI-generated synthesis risks eroding critical engagement and intellectual independence. 4. Reconfiguration of scholarly authority Metrics, rankings, and algorithmic recommendations reshape perceptions of expertise and credibility. 5. Hidden curricula of evaluation AI-mediated knowledge systems implicitly train scholars to value certain outputs and behaviours without explicit reflection. Addressing these challenges requires renewed attention to critical thinking as a scholarly practice, not merely a methodological skill.
Contributors
• Mr. Francis P. Crawley (Course Chair), GCPA & SIDCER, Leuven, Belgium (fpc@gcpalliance.org) • Dr. Pascal Mwakio (Course Co-Chair), KEMRI Bioethics Training Initiative, Nairobi & Bioethics Program Lead at AMREF International University, Nairobi, Kenya • Dr. Adriana Baravalle (Course Co-Chair), Department of Computer Science & Department of Artificial Intelligence, Universidad Austral, Pilar Campus, Argentina • Dr. Amos Mbugua (Course Co-Instructor), Chair, Ethics Committee, Jomo Kenyatta University of Agriculture & Technology (JKUAT), Juja, Kenya • Dr. Lili Zhang(Course Co-Instructor), Computer Network Information Center, Chinese Academy of Sciences, Beijing, China • Dr. Rodrigo Roa (Course Co-Instructor), Executive Director, Data Observatory, Providencia, Santiago, Chile • Dr. Vina Vaswani (Course Co Instructor), Centre for Ethics, Yenepova (Deemed to be University), Mangaluru, India • Professor Mara de Souza Freitas (Course Co-Instructor), Director, Institute of Bioethics, Universidade Católica Portuguesa (Catholic University of Portugal), Lisbon, Portugal • Professor Perihan Elif Ekmekçi (Course Co-Instructor), TOBB University, Ankara, Turkey • Professor Valery Sokolchik (Course Co-Instructor), Belarus Academy of Sciences, Minsk, Belarus • Bert Seghers (Course Guest Presenter), Flemish Commission for Research Integrity (VCWI) & President, European Network of Research Integrity Offices (ENRIO), Belgium:
Target Audience
This course is designed for researchers, educators, librarians, editors, publishers, research assessors, and policy-engaged scholars who are already familiar with open
scholarship and research integrity issues and wish to deepen their understanding of how AI reshapes scholarly thinking and responsibility. Participants should be interested not only in how AI is used in scholarship, but in how it transforms the conditions under which scholarly judgement is exercised.
Engagement
The course emphasizes active participation, reflective discussion, and peer-to-peer exchange. Recognizing the expertise of participants, sessions are structured to encourage critical dialogue rather than one-way instruction. Online learning modalities • Interactive live sessions Online modules will include live presentations paired with interactive components such as polls, Q&A sessions, and open discussions to encourage active participation. • Collaborative group work Participants will work in small breakout groups to tackle epistemic case analysis and analyze real-world scenarios related to AI in scholarly communication. • Discussion forums An online platform will host asynchronous discussions, allowing participants to share ideas, resources, and insights at their convenience, fostering ongoing dialogue between sessions. • Practical exercises Online tools will be used to simulate AI-assisted knowledge generation workflows, enabling participants to critically evaluate the capabilities and limitations of these technologies in a hands-on manner. In-person nodes • Focused workshops In-person sessions will be organized around specific challenges, such as redefining knowledge frameworks and developing responsible epistemic practices and frameworks, allowing participants to collaborate intensively in smaller, focused groups. • Peer-to-peer learning Participants will share their experiences and insights, contributing to a collective understanding of best practices for integrating critical thinking and responsible epistemics into scholarly communication. • Co-creation sessions In-person activities will prioritize identification of AI challenges in generating knowledge through shared reflection, epistemic dialogue, and comparative analysis ideas, with participants refining ideas generated during the online modules and tailoring them to different disciplines, regions, or research contexts. • Networking opportunities Informal discussions and social events will facilitate connections among participants, fostering partnerships and collaborations beyond the course. By combining the flexibility and reach of online learning with the depth of in person engagement, this distributed course ensures that participants are actively involved in shaping discussions, applying new knowledge, and contributing to high impact outcomes. The strategy promotes a balance of theoretical exploration and practical application, enabling participants to leave with actionable insights, valuable
networks, and tangible contributions to the global discourse on critical thinking, epistemic responsibility, and open scholarship in the age of AI.
Main Topics
The course covers a range of critical topics aimed at addressing the complex implications of AI in knowledge generation and scholarly communication on authorship as ‘a site of knowing’. These topics are designed to help early career, seasoned scholars, and professional scientists understand how AI is transforming traditional learning and science practices, redefining scholarly communications are developed, and introducing new challenges to research integrity. Through discussions, epistemic case analyses, and practical exercises, participants will explore these themes in depth, gaining actionable insights and strategies to address the evolving landscape of AI in academia and science generally. • Critical thinking under AI-mediated knowledge production • Epistemic authority and responsibility in automated environments • Human judgement and the limits of algorithmic evaluation • AI, research assessment, and the shaping of scholarly values • Transparency, humility, and trust in open scholarship ecosystems • Sustaining human agency in AI-enhanced research infrastructures By addressing these topics, the course equips participants with the knowledge and tools necessary to navigate the transformative role of AI in scholarly knowledge generation and communication. The focus on practical and high-level outcomes ensures that attendees leave not only with a deeper understanding of the challenges and opportunities posed by AI for scholarly knowledge but also with concrete strategies to lead in shaping the future of responsible scholarly practices.
E04: Assessing Artificial Intelligence Tools for Scholarly Use
Class Time: 9:00 AM – 12:00 PM (Pacific Time – UTC-7)
Instructor Abirami Muthukumar
Affiliation University of British Columbia
Job Title Humanities & Social Sciences Librarian (Digital Scholarship)
Course Abstract
This course will introduce participants to core concepts in how to vet Artificial Intelligence (AI) tools for scholarly use. Digital tools used in scholarly work are gaining new layers of complexity due to the increasing prevalence of AI and machine learning methods integrated into everyday features. It is increasingly important to assess tools not only in terms of their utility but also in terms of current Copyright and publishing standards, security, reproducibility, and longevity. This assessment is made even more challenging when AI features are a “black box” which refers to a system that obscures what it is doing and how it works from a users perspective. Black box AI can inhibit the process of scholarly inquiry or publishing through the obfuscation of methods that have traditionally been transparent. This course will introduce participants to the core considerations in vetting AI tools with specific consideration given to mapping tool features to the requirements of a project. We will cover the concept of the black box and ways to gain insight into the decision making happening within the tool even when it is obscured by design. We will also address how to identify whether a tool has a black box
even when it is not immediately clear and the major pitfalls to look out for. We will explore several of the most common AI tools used in academia for tasks ranging from summarization and note-taking, annotation and distant reading, to the generation of outputs with the intention of publication. Participants will be lead through sample activities to assess Microsoft CoPilot and Open AI’s ChatGPT using a rubric we provide. They will then proceed to pair a project with a scholarly-focused AI based tool from a list and assess it using the rubric in a hands on activity. If participants have a project which would benefit from the use of AI tools they are welcome to use it as an example for this part of the course, a project is not required. At the end of this course participants will be able to: – identify the different types of AI systems – define what a black box AI is – articulate common pitfalls with AI tools in relation to scholarly work including reproducibility and scholarly integrity – apply a rubric for the assessment of common AI tools in relation of their utility, safety, and reproducibility for scholarly work – identify and apply strategies for the mediation of common pitfalls – effectively cite and describe the use of an AI tool in scholarly output While AI tools carry significant risks to academic integrity there are ways to mediate black box aspects and support greater reproducibility in the final scholarly outputs. Our goal is to empower participants with the core skills needed to counter these issues and make use of a powerful new suite of possibilities.
Contributors
Ekatarina (Eka) Grgurić, Digital Scholarship Librarian, University of British Columbia (co instructor), Erin Fields, Open Education & Scholarly Communications Librarian, University of British Columbia (co-instructor)
Target Audience
Researchers, administrators / decision makers who choose tools to purchase, librarians, support staff.
Engagement
Our course will have two major activities as well as several moments where participants will be able to engage through active learning. We will begin the course with a virtual whiteboard activity to identify tools and project types attendees are arriving with interest in. The lecture portions of our course will have several Q&A opportunities as well as active learning elements such as a Zoom quiz and a voting activity that will use emojis. The two main activities will run as follows: Activity 1 – participants will follow along with a rubric that the instructors provide while assessing Microsoft Copilot and Open AI’s ChatGPT with consideration of project requirements for a fictional research project. For this activity attendees will complete a pre-filled rubric with significant instructor guidance. Activity 2 – participants will suggest example projects and AI tools of interest in the chat and instructors will organize the suggestions on a virtual whiteboard, to lower the barrier to participation instructors will also provide a few example projects and example AI tools to investigate. Participants will then split into breakout rooms and do a rudimentary assessment of one tool with consideration of a few project requirements using the same rubric as before, this time completely empty. Groups will come back together and report their findings at the end of this activity.
Main Topics
Artificial Intelligence Tools, Computational and Research Reproducibility, Mitigating Black Boxes, Research Integrity
E05: Library Publishing and Global Scholarly Communication
Class Time: 9:00 AM – 11:00 AM (Pacific Time – UTC-7)
Instructor Jonathan Grunert
Affiliation University at Buffalo
Job Title Scholarly Publishing Librarian
Course Abstract
Shifting landscapes of academic publishing, open access initiatives, transformative agreements, and questionable publishing practices have all contributed to an evolution in the role libraries play within academic institutions. Library publishing programs emerged from this challenge as a viable alternative to traditional publishing outlets, being more affordable than scholarship produced for and published by legacy publishers. In a global scholarly communication ecosystem, library publishing programs can have a broad impact, especially when scholarly communication professionals utilize those services to expand the reach of open research. Initially, this course examines the current state of library publishing, including its functions, how it exists within a scholarly communication context, and its global reach. We will identify potential uses of library publishing for scholarly communication professionals, including important discussions of personnel who are or can be responsible for library publishing, as well as practices for engaging with library publishing both in and out of home institutions. This course draws from personal experience in library publishing as well as readings that detail recent innovations and research in that arena. Participants in this course do not need to have publishing experience (library or otherwise), as the course functions both as an introduction to library publishing as well as an exploration of significant topics in library publishing. By the end of the course, participants will have a greater understanding of
library publishing in a global scholar communication context, and they will have a plan for increased engagement with library publishing at their home institutions.
Course Outline
Session One: Pre-class reading on library publishing writ large. In-class, examine publishing in a library context, differentiating between library publishing and university press. Discuss why library publishing is (or may be) a scholarly communication concern.
Session Two: Pre-class work includes identifying and exploring a library publishing program at a similar institution to your own. In-class, extend library publishing to a global context— what can library publishing offer for a global scholarly communication ecosystem? Identify library publishing personnel responsibilities.
Session Three: Pre-class work identifying who at your institution already does some of these responsibilities. In-class, connecting the library publishing responsibilities to personnel, both connected to and disconnected from scholarly communication roles. Generate plans for engaging with library publishing at your home institutions, whether starting a new program or expanding it.
Target Audience
Researchers, administrators, librarians, publishers The course explores the global reach of library publishing.
Engagement
Planned activities include identifying and analyzing library publishing programs, evaluating personnel responsibilities, and planning future engagement.
Main Topics
Library publishing, open access, predatory publishing
E06: Assessing Institutional Research Data Using OpenAlex and R
Class Time: 9:00 AM – 11:00 AM (Pacific Time – UTC-7)
Instructor Teresa Schultz
Affiliation University of Nevada, Reno
Job Title Scholarly Communications & Social Sciences Librarian
Course Abstract
OpenAex, an emerging open source for research data, offers an intriguing alternative to expensive options such as InCites and Dimensions. However, working with the data at a large and in-depth level requires more technological know-how than those who work in
scholarly communications are trained in. This workshop will walk users through how to connect to OpenAlex’s API using an R package and provide them with ready-made code and detailed instructions on how to access different types of data through OpenAlex and then transform and analyze it to learn things like: – Where an institution’s researchers are publishing – Where their coauthors come from – What is their participation like in open access – Who is paying to publish open access, with and without funding However, because it is likely that individual participants will want to explore the data in various ways specific to only them, the workshop will also explore the use of generative AI to build on the provided code to further explore and use the data, such as having AI assist in creating data visuals and reproducible reports based on reproducible data. By providing a basic foundation of R while also exploring the use of AI to help code, the workshop will provide participants with valuable experience in coding that they can then use to better understand and control the use of AI in aiding them further. The workshop will also discuss the advantages and disadvantages of using OpenAlex as a datasource, especially in comparison to other tools, along with a background in what research data from these sources can actually tell us and what they
can’t. Participants are not expected to be familiar with R or RStudio, although participants should not expect to be experts in R by the end. Participants will be expected to do some work before the workshop, and there will be some homework during the workshop as well.
Outline:
Day 1 will focus on an overview of OpenAlex, including:
• What data it includes and from where
• How data is structured in OpenAlex
• Known issues with OpenAlex data
• How to gather data from OpenAlex
The second half of the first day will then introduce participants to R and RStudio, giving an overview of how R works, as well as how to use R to collect data from the OpenAlex API through the openalexR package.
Day 2 will then learn how to clean, process, and analyze the data and conduct initial analysis.
Day 3 will focus on refining the data for analysis of OA participation rates and how to then create data graphics in RStudio. Participants will also learn how they can use generative AI to go beyond the provided code, as well as the pros and cons they need to consider when doing so. Contributors
Co-instructor Carlos Ramirez-Reyes, Data Services Coordinator, University of Nevada, Reno
Target Audience
Librarians, researchers interested in bibliometrics, and anyone involved in research assessment
Engagement
The first day will be mostly lecture, which will include interactive online polling through Mentimeter. However, all of days 2 and 3 (and even part of day 1) will provide participants a chance to actually use code in RStudio as well as pull data from OpenAlex and use generative AI to create code they can then run in RStudio. Participants will follow along as instructors demonstrate how to run code and what they should see and then do so themselves.
Main Topics
Research metrics, bibliometrics, institutional impact
Requirements
Participants will need computers that can run R and RStudio. These do not need to be advanced computers. The instructors are able to run them on personal computers that are five years old.
L08: Governing Thought in the Age of AI: Diagnosing and Countering Epistemic Failure Modes in AI-Engaged Scholarship
Class Time: 4:00 PM – 6:00 PM (Pacific Time – UTC-7)
Instructor Shana Pote
Affiliation Aurorae Group (with academic affiliation: Fielding Graduate University) Job Title Executive Coach, Educator, and Doctoral Researcher
Course Abstract
Generative AI is now embedded in scholarly work, yet most conversations focus on ethics, policy, or tools rather than on how AI reshapes scholarly thinking itself. In
practice, scholars engage generative systems in multiple ways, such as assistants, accelerators, surrogate thinkers, or dialogic interlocutors, and each mode introduces distinct epistemic failure modes. These include judgment abdication, premature convergence, and authority displacement, as well as more emergent risks that arise when AI is used dialogically to accelerate learning and conceptual development. This course presents a systems-level framework for thinking about thinking with generative AI. Participants will examine four modes of AI–human collaboration, identify the characteristic epistemic failures associated with each mode, and learn how different forms of AI use demand different forms of critical governance. Particular attention is given to dialogic engagement with AI, which can strengthen reasoning and expand conceptual reach while also introducing new risks such as assumed shared foundations, false fluency, and conceptual smuggling. Rather than offering tools or prescriptions, the course focuses on diagnostic and governing practices for actively countering epistemic failure modes. Participants will learn how to recognize epistemic failure modes as they arise and how to select mode-appropriate counter-practices, including reasserting evaluative criteria and authorship, deliberate divergence and counterfactual exploration, role clarification to resist authority displacement, and epistemic interruption in service of conceptual grounding. The emphasis throughout is on retaining judgment, grounding, and authorship as generative systems increasingly shape scholarly dialogue, learning, and knowledge production. Designed for researchers, librarians, educators, and scholarly communication professionals, this course equips participants to think more clearly about how they think with AI, and to teach, model, and support epistemically responsible practice across varying levels of AI sophistication. By the end of this course, participants will be able to: 1. Differentiate four distinct modes of thinking with generative AI and explain how each reshapes scholarly judgment and responsibility. 2. Identify epistemic failure modes associated with different modes of AI use, including judgment abdication, premature convergence, authority displacement, assumed shared foundations, and false fluency. 3. Diagnose when language, momentum, or fluency have outpaced understanding in AI-engaged scholarly dialogue. 4. Select and apply mode-appropriate counter-practices to govern epistemic risk, rather than relying on universal or tool based fixes. 5. Use epistemic interruption and conceptual grounding where warranted to clarify unfamiliar constructs and restore epistemic alignment. 6. Articulate how these diagnostic and governing practices can be taught, modeled, or supported in scholarly and educational contexts.
Course Outline
The course will be delivered over three 2-hour synchronous sessions during the FSCI week, combining short presentations, guided discussion, and applied diagnostic and counter-practice activities. Sessions are designed to build from system-level framing to recognition of epistemic failure modes and finally to the selection and teaching of appropriate counter-practices. Collaborative discussion and reflection will be emphasized over tool demonstrations.
Session 1: How Scholars Think With AI Introduces four modes of thinking with generative AI and examines how each reshapes scholarly reasoning, responsibility, and authorship. Participants identify common epistemic failure modes such as judgment abdication, premature convergence, and authority displacement, and reflect on how these appear in their own contexts. Activities: • Guided self-mapping exercise where participants identify how they currently engage generative AI across the four modes • Small-group or whole-group discussion diagnosing which epistemic failure modes appear most often in their own contexts
Session 2: Dialogic AI Use and Emergent Failure Modes Focuses on dialogic engagement with AI as a distinct mode of collaboration that both strengthens reasoning and introduces new risks. Participants examine failure modes such as assumed shared foundations, false fluency, and conceptual smuggling through case-based discussion. Activities: • Case-based analysis of short AI–human interaction scenarios to identify assumed shared foundations, false fluency, or conceptual smuggling • Facilitated discussion comparing how dialogic use differs epistemically from assistant or accelerator use
Session 3: Recognizing and Countering Epistemic Failure Modes Centers on learning to recognize epistemic failure modes as they arise and to select appropriate counter practices. Participants practice selecting and applying strategies such as reasserting evaluative criteria, deliberate divergence, role clarification, epistemic interruption, and conceptual grounding, with attention to how these practices can be taught or modeled for others. Activities: • Mode–failure–counter mapping exercise where participants match specific failure modes to appropriate counter-practices • Structured reflection on how these recognitions and responses could be modeled or taught in participants’ own scholarly or instructional settings
Student Prerequisites
This course is intended for intermediate to advanced learners who are already engaging with generative AI in scholarly, instructional, or research support contexts and want to think more critically about how AI reshapes judgment, authorship, and knowledge production. It is well suited for researchers, faculty, librarians, scholarly communication professionals, research administrators, and others involved in supporting or evaluating scholarly work. Participants do not need technical expertise in AI or computer science. However, the course assumes a basic familiarity with academic or research workflows and a willingness to reflect on one’s own practices and assumptions. The course is discussion-driven and diagnostic in nature, and will be most valuable to participants who are comfortable engaging in reflective, conceptually oriented conversations rather
than tool-based training. No preparatory work is required. Participants are encouraged to bring examples from their own experience using generative AI in research, teaching, publishing, or scholarly communication, but these can be informal and do not require advance preparation.
Target Audience
This course is designed for researchers, librarians, educators, research administrators, and scholarly communication professionals who are engaging with or supporting the use of generative AI in research, teaching, or knowledge production. No technical background in AI is required.
Engagement
The course will be delivered over three synchronous sessions during the FSCI week, combining short presentations, guided discussion, and applied diagnostic activities. Sessions are designed to build from system-level framing to recognition of epistemic failure modes and finally to the selection and teaching of appropriate counter-practices. Collaborative discussion and reflection will be emphasized over tool demonstrations.
Main Topics
Critical thinking and epistemic judgment in scholarly work Generative AI and scholarly communication Human–AI collaboration in research and knowledge production Authorship, attribution, and responsibility under AI-assisted scholarship Research integrity beyond compliance and policy Epistemic risks and failure modes in AI-engaged research practices Evaluation and legitimacy of AI-engaged scholarly work Scholarly writing and reasoning under technological acceleration Sociotechnical perspectives on AI and knowledge work Teaching and supporting critical AI use in research and academic contexts
L09: Open as Antidote – addressing the big challenges facing science
Class Time: 5:00 PM – 6:30 PM (Pacific Time – UTC-7)
Instructor Danny Kingsley
Affiliation Deakin University
Job Title Director, Library Services (Information). I am also a volunteer Ambassador for the Directory of Open Access Books (DOAB)
Course Abstract
An antidote is something that acts against the effects of a poison or disease. It is not too much of a stretch to consider what has been happening to science in the past couple of years as a form of poison. But we were not in a utopia prior to the current geopolitical situation. The research ecosystem has been under considerable pressure for a long time. Many aspects of research operate in isolation from each other yet are part of an interdependent whole. Areas such as research culture, research assessment, open scholarship, research integrity, research support, research infrastructure and research impact can be managed by completely different agents within a research institution (if at all). The training we offer all participants in the research endeavour does not currently take a holistic view, which makes effective decision-making and positive change deeply challenging. This course is an attempt to articulate the interdependent connections across the ecosystem and identify areas where change might be possible. In this course classes will be 1.5 hours long, with three classes and will build on some pre-work for each class. The classes will combine direct instruction with discussion, small group work and whole group activity.
Course Topics
This course will be presented over three days for 90 minutes each day and will cover these topics:
● Topic 1 – How bad is it really? A dive into the accelerating increase in poor research practice, paper mills, fraud, retractions … (yes this is the depressing lesson!)
What has been happening over the past 20 years in terms of changing research assessment, the ownership of research infrastructure, and the rise of the Open movement
● Topic 2 – The elephant in the room Considering the twin challenges for research – the rise of AI and the geopolitical environment. Where are we if there is no trust in research?
● Topic 3 – The open antidote A celebration of – and critical look at – initiatives that are shifting the dial. The common thread is the connection across different aspects of the ecosystem. We will finish by looking back to see forward.
Target Audience
All participants in the research space – researchers, research administrators, librarians, third space professionals – everyone. It is intended to be interactive and constructive where all participants can contribute to the process.
Engagement
I will have pre-recorded presentations as pre-work, so the chalk and talk aspect is covered off asynchronously. The session each have a mixture of quizzes (Mentimeter) and discussion points. I only had a small group last year so it took a bit to get discussions going but they engaged in the chat.
Main Topics
Questionable research practices, fraud, paper mills, predatory publishing, research infrastructure, research integrity, open access, data sharing, open infrastructure, research assessment (all of it really).
L10: Web Archiving, Reference Rot, and Research Integrity
Class Time: 5:00 PM – 6:30 PM (Pacific Time – UTC-7)
Instructor Rosario Rogel-Salazar
Affiliation Universidad Autónoma del Estado de México
Job Title Web Archiving, Reference Rot, and Research Integrity
Course Abstract
The integrity of scholarly communication increasingly depends on digital resources that are volatile, platform-dependent, and now frequently mediated by artificial intelligence systems. While reference rot and content drift have long threatened the reproducibility of research, the rise of AI-generated content, algorithmically curated web environments, and dynamic knowledge graphs introduces new preservation challenges.
This course examines web archiving as a critical scholarly infrastructure in the age of AI. Participants will explore how reference rot, platform decay, and algorithmic transformations affect research integrity, open science practices, and FAIR-aligned scholarship. The course combines conceptual discussion with hands-on training in web archiving tools, while critically assessing the limits of preservation when content is dynamically generated, personalized, or AI-mediated.
Designed for researchers, librarians, editors, and scholarly communication professionals, the course emphasizes practical strategies, editorial workflows, and policy-aware approaches to preserving the digital scholarly record under conditions of technological acceleration.
Outline / Course Structure & Schedule
Total duration: 4.5+ hours
Format: 3 live Zoom sessions, plus optional asynchronous activities
Session 1 – The problem has changed: from link rot to algorithmic volatility • Reference rot and content drift: what we already know.
• What changes when content is AI-generated, continuously updated, or personalized?
• Case examples: disappearing policy documents, rewritten webpages, AI-assisted research outputs.
• Discussion: reproducibility in an AI-shaped web.
Session 2 – Web Archiving as Scholarly Infrastructure
• Overview of web archiving ecosystems and tools.
• Hands-on demonstrations: Wayback Machine, Perma.cc, Webrecorder, Archive-It. • Strengths and limitations: what can and cannot be preserved.
• FAIR, CARE, and beyond: where preservation frameworks fall short. Session 3 – From Tools to Practice: Editorial, Research, and Policy Strategies • Preventing reference rot at the source: author and editor responsibilities. • Web archiving in journal workflows, repositories, and open peer review. • Preservation ethics: consent, authorship, AI outputs, and institutional accountability. • Group exercise: designing a preservation strategy for a real scholarly context.
Contributors
Alan Colín-Arce, guest presenter
Target Audience
This course is designed for researchers, scholarly communication librarians, journal editors and publishers, and research support professionals working with web-based sources. It will also appeal to open science and research integrity communities, as well as practitioners involved in digital preservation, web archiving, and FAIR-aligned scholarly workflows.
Learning Objectives
By the end of the course, participants will be able to:
1. Explain how reference rot, content drift, and AI-mediated web content affect research transparency and reproducibility.
2. Critically assess the limits of FAIR principles when applied to non-static, AI generated, or platform-hosted web resources.
3. Use web archiving tools (Wayback Machine, Perma.cc, Webrecorder, Archive-It) strategically within scholarly and editorial workflows.
4. Identify emerging preservation risks associated with generative AI, algorithmic personalization, and platform dependency.
5. Design discipline- and role-specific strategies (author, editor, librarian, policymaker) to mitigate reference loss.
Engagement
The course will combine short interactive lectures with hands-on activities using web archiving tools. Live sessions will include guided demonstrations and, when participant numbers allow, small-group discussions to encourage peer learning and exchange across roles and disciplines. Light asynchronous activities, such as brief reflections or resource sharing, may be used between sessions to support continued engagement and practical application in participants’ own contexts.
Participant Activities
• Hands-on archiving of selected web resources.
• Critical analysis of a “fragile” reference list from participants’ own work. • Collaborative design of a preservation protocol for an AI-affected research scenario. • Optional asynchronous reflection or shared resource board.
Target Audience
• Scholarly communication librarians
• Journal editors and publishers
• Researchers using web-based and AI-assisted sources
• Open science practitioners
• Research data managers and policy advisors
Main Topics
– Scholarly communication infrastructure – Research integrity and reproducibility – Reference rot and content drift – Web archiving and digital preservation – FAIR-aligned scholarly practices – Use of web-based sources in research and publishing – Editorial and publishing workflows – Open science and open research practices – Archived web content as research data
Requirements
The technical requirements for this course are minimal. Participants will need a stable internet connection, access to Zoom, and a computer with a modern web browser. No special equipment is required. All tools demonstrated in the course are web-based and free to access; no software installation is required. Participants should have basic digital literacy and be comfortable navigating online resources. Teaching aids will include screen sharing for live demonstrations and optional collaborative tools (e.g., shared documents or boards), all accessible through a web browser.
