The Power of Open

in an AI Era

How Open Practices Help Us Adapt and How AI Can Help Open Grow

Anna Mills, OpenCon Ohio Keynote, May 15, 2026

​

Welcome! Please share your name, where you work, your role, and your favorite workday drink. Here’s the plan:

• How Open Practices Help Us Adapt to AI
• Poll
• How AI Can Help Open Grow
• Poll and Q&A

​

Slides, open for comments:�https://link.annarmills.com/OpenCon

Part 1: How Open Practices Help Us Adapt to AI

In 2023, Maha Bali, Lance Eaton, and I wrote an article about Open Educational Practices and AI

How do we respond to generative AI in education? Open educational practices give us a framework for an ongoing process

Commentary, Journal of Applied Learning and Teaching

Published June 11, 2023

Anna Mills, Maha Bali. Lance Eaton+

​

What are Open Educational Practices?

“Cheryl Ann Hodgkinson-Williams and Henry Trotter (2018) point out that open educational practices can include curation and distribution of resources, facilitated by crowdsourcing and open peer review.

…[O]pen educational practices focus on forms of community that cross institutional boundaries, disciplinary silos, and national borders.” (Mills et al, 2023).

​

In the paper, we emphasized less formal, nontraditional kinds of professional development.

• Personal Learning Networks that draw on cross-institutional social media, groups, and blogs as open learning space/sharing space
• Sharing rough, being vulnerable
• Crowdsourcing curation and creation

​

All these help us be more flexible, vulnerable, curious, and less scared. They help us develop more nuanced approaches and keep building.

I would argue that we saw a flowering of all those practices and that we need them now more than ever. The shock of AI isn’t over.

• It's a shock that comes in waves. AI capacities change and increase:
• Systems now browse and work with real sources. They are more sophisticated in terms of the kind of reasoning they can simulate. They work with and analyze all kinds of media: slide decks, images, video
• We're seeing the beginning of an agentic AI era where AI uses tools well enough to function as an imperfect personal assistant.

Social media as a learning and community space

• We all know the concerns and complaints about social media as polarizing, commercializing space full of bots and trolls. They’re not groundless.
• But there's also a tale of open practices on social media spaces like LinkedIn and BlueSky: a tale of resilience and community and collaboration.
• Contrary to reputation, there is a lot of curiosity, mutual aid, and stimulating discussion.

I sought feedback on a custom bot on LinkedIn recently:

There were so many engaged responses: pushing back, sharing parallel efforts, and coaching me on refinements.

For example, Remi Kalir said, “stepping back…do I/you, as an educator, design this type of assistive bot for students or should I/you teach my students to query any LLM to receive this type of assistance?

My post came from uncertainty, and it struck me that more people responded than to my usual opinion or resource posts.

Jason Gulya celebrates how on his LinkedIn community leads him to question his ideas

“Over the past few months, I've been writing a book on teaching process in The Age of AI. As I write and share my ideas there, my [LinkedIn] community has really engaged with me. They've pushed against my assumptions, and show me (again and again) just how complex questions about AI and education are.

— Jason Gulya, Problems with "Process Over Product" Part 1, October 17, 2025

Private groups and blogs with commenting

• AI in Education listserv (3,000+ members, very active)
• Laura Dumin's Facebook group Higher ed discussions of AI writing & use 10,000 members and many posts each day. Anonymous posts allowed.
• In both groups, people share frustrations and uncertainties, ask questions, and share resources or experiments.
• Bloggers sharing about their evolving thinking on AI in education: Liza Long, Joel Gladd, Maha Bali, Lance Eaton, Marc Watkins, Mike Caulfield, Leon Furze, Ethan Mollick, Josh Brake, David Wiley, Karen Costa, for a start…

Professional development sharing

• The AI Faculty Cohort Programs Collection developed by Lance Eaton at the Northeastern University Center for Teaching and Learning, based on a survey shared on social media and lists. “This is an exploratory field scan. There are no “best practices” presumed here. The goal is to surface patterns, variations, and emerging approaches in order to better understand what institutions are trying, what questions are shaping this work, and where there may be opportunities for shared learning. “
• I see constant free webinars and workshops announced on LinkedIn. Many institutions have put open licenses on their faculty-facing modules. See my list of Ongoing Sources of AI Professional Development:
• Claude-generated draft spreadsheet of faculty-facing OER about AI

Student-facing materials on AI

Claude-generated spreadsheet of student-facing OER materials

I put out a draft of my OER text AI and College Writing: An Orientation as a Google Doc with a “Request for comments"

“I would appreciate feedback from both teachers and students on this draft! (I’m inspired by the open peer review process for Remi Kalir’s book Annotation, as described by Troy Hicks.) Please join in by leaving public comments on this Google Doc or by emailing me at info@howargumentswork.org to suggest a way you'd like to contribute.”

Sharing the rough version made for more engagement

• I put my call for feedback out on social media. Some people started using the text even though it was still a draft.
• The draft got some 500 visits. Lisa Sperber, Maha Bali, David McHugh, Tamara Tate, Ryan Hitch, Ted Stresen-Reuter and quite a few others made very helpful comments.

Policy and Pedagogy sharing

• Lance Eaton's AI syllabus policy repository
• AIpedagogy.org, a great collection hosted by the metaLAB (at) Harvard.
• Jon Ippolito’s LearnwithAI.org, hosted at the University of Maine.

Sometimes crowdsourced collaboration leads to apps

Jon Ippolito’s What Uses More project allows students and instructors to compare and discuss relative environmental impacts of digital activities. Inspired and informed in part by discussions and contributions from the AI in Education listserv.

Collections of prompts for higher ed

• Jose Antonio Bowen’s list of prompts described in Teaching with AI
• The Peer & AI Review + Reflection prompts
• PapyrusAI prompt library
• Lance Eaton's prompt library
• David Wiley's Generative Textbooks project is a kind of prompt library for learning activities related to learning outcomes/existing content:
• Prompt Patterns from Vanderbilt University
• Maastrict University AI Prompt Library
• Ethan Mollick and LIllach Mollick’s prompt library for education

Mike Caulfield argues that now more than ever, we can shape chatbot output according to our teaching expertise

“ As the systems are able to process more complex instruction sets, people who…can break that down for students suddenly have an advantage. They can now break it down for a machine…

” – Mike Caulfield, “How Prompting Is Evolving,” July 2, 2025

On some custom bot platforms, you can view and share prompts

PlayLab is a nonprofit whose purpose is to allow educators to make custom bots and remix each other’s. Once you sign in, you can see the prompts for many bots if you remix them. If you aren’t able to get into the training, you can start out in the workspace I created for educators.�

​

An open process: the Mike Caulfield example

Beginning March 25th, 2025, Caulfield shared some 17 Substack articles and many videos about the evolution of the megaprompt he first called SIFT Toolbox.

He pulled out and explained the purpose of particular parts of the prompts, shared his testing process, explained how he was updating the prompt at each stage, and gave examples of outputs.

Regular readers tested it, commented on the Substack, and offered suggestions and troubleshooting.

Caulfield’s expertise as an educator informs the prompt. He co-wrote the book Verified and came up with the SIFT method for verification, widely taught in higher ed.

Caulfield shared the text of the whole 3,000+ word prompt (this is the beginning)

He put up a website with the updated prompt along with text and video instructions and examples. I learned prompting techniques by seeing them in action.

The Peer & AI Review + Reflection Project (PAIRR) Curriculum Committee tested a writing feedback prompt

• PAIRR was founded at UC Davis and is supported by the California Education Learning Lab
• The PAIRR grant leadership includes faculty from 8 public colleges and universities in California. We train cohorts of faculty at each college to use our methodology and initial AI feedback prompts as part of a study.

We’ve shared the Peer & AI Review + Reflection packet of OER assignment materials

• Feedback prompts we’ve written and tested
• Scaffolding that helps students chat back to the feedback
• Reflection questions to encourage skepticism and sense of agency

​

Public sharing helped us improve our prompts: We revised them after feedback gathered at a public webinar in July 2025.

We shared links to the full text of a reader-response prompt and a criteria-focused prompt and explained their different approaches.

Then we asked participants to read two contrasting samples of feedback and complete a poll on which they preferred and why.

We’re working to encourage adoption and adaptation

• Feli Hartung and Ruth Poproski at the Center for Teaching and Learning at the University of Georgia adopted and adapted PAIRR materials and reshared them.
• Jason Gulya wrote about his adaptation of PAIRR on his Substack
• 4,000+ visits to our OER materials packet
• We made a Padlet and a Google Group where people can comment and share adaptations and we are active on social media.

Agentic AI: the next shock or wave, starting yesterday

• It’s only within the last nine months or so that AI systems can use tools pretty effectively. This creates new academic integrity concerns around automated course completion, and it also means educators might benefit from agentic AI assistance. But it carries privacy and security risks. Capacities are changing rapidy.
• Social media and blogs are the place to learn and share experiments.

Agentic AI uses tools, including the cursor

A few examples of open sharing about agentic AI

• Liza Long and Joel Gladd have shared brilliant examples of workflows involving Claude Code and Codex. (My summer goal is to try at least one of these.) Don’t miss Liza’s plenary today about her agentic workflow for OER accessibility!
• She has written about the overwhelm associated with vibe coding in a very vulnerable and helpful way.
• Leon Furze has shared A Taxonomy of Agentic AI
• Teddy Svoronos reflects on uses in his statistics class and his internal debates
• I shared slides on agentic AI uses and abuses and a LinkedIn skill for Claude

Poll

• Where do you turn or hope to turn to learn about AI in Edu?
• What open resource or sharing space do you wish for that you haven't found yet?

​

• Go to https://www.menti.com/alwd1nt5hu4d (link in the chat) or to Menti.com and enter 6509 9225

Part 2: How AI Can Help Open Grow

It could undermine what we're trying to do, or it could support it

• In my mind, the core of OER/open pedagogy/open practices is human relationship: teaching not for profit but out of a desire to communicate, share, and help.
• I think of each textbook as a letter to students, based on my and my collaborators teaching experience and studies. We’re saying, “Here's what we find valuable; what do you make of it?”
• If the instructional materials don’t come from humans, we could lose that message of care.

Students may perceive neglect and hypocrisy when instructional materials are made with AI

The Guardian, November 2025

[O]h, how the tables have turned. Now students are complaining on sites like Rate My Professors about their instructors' overreliance on A.I.." — Kashmir Hill, The New York Times, May 14, 2025

But we can set limits.

“Integrating GenAI into OER workflows, however, is not simply a matter of adding a new tool. It raises questions about authorship, control, instructional intent, and pedagogical value. Using GenAI effectively in OER creation requires educators to define its scope, determine where it meaningfully supports learners, and assess where a human-centric approach should be prioritized.” — Lance Eaton and Larry Davis, Expanding OER with GenAI,” Educause, May 7, 2026

Aras Bozkurt on enhancement

“It's not a matter of replacing human authors or educators, but rather, of enhancing their capabilities. In the case of generative AI, its outputs can be harnessed to create more diverse, expansive, and inclusive open educational resources, transforming not just how content is created but also how it's used and shared.”

​

— Aras Bozkurt, Bozkurt, A. (2023). Generative AI, Synthetic Contents, Open Educational Resources (OER), and Open Educational Practices (OEP): A New Front in the Openness Landscape. Open Praxis, 15(3), pp. 178–184. DOI: https://doi.org/10.55982/openpraxis.15.3.579

Lance Eaton's framework for using AI with OER (2025) and the followup Educause article “Expanding OER with GenAI” (2026)

The one thing I don’t do from Lance’s framework is “Co-Create”

• I’m currently not using AI for ideas, drafts, or even outlines. I try to draw a line so I can tell students this is human guidance, ideas, and writing.

​

• It gets tricky. Gray areas include
• chats with AI where I want it to draw out my thinking
• worksheets or slides that summarize or reword the original OER content

My preferred uses of AI to support OER

• Curation research assistance. For lots of purposes: finding OER texts to adopt, finding material to consider remixing, researching places to share the OER. See Easier Strategies for Finding and Evaluating OER Resources with and without AI.
• To convert to another format. I.e. make slides based on this OER chapter. Customize to topics of interest. Offer different learning modalities. Tool using agentic AI is helpful here. See Plenary 2: Accessibility by Design: Agentic AI for Pressbook Accessibility with Liza Long.
• To co-create supplementary materials like problem sets and examples.
• For feedback on my OER drafts including research and fact checking.

A sample agentic workflow

• I started with Gemini in Chrome editing Google Slides and wasn't happy with the performance.
• I switched to Claude Cowork which used Claude in Chrome, Google connectors, and PowerPoint on Google Drive. It asked clarifying questions and narrated its progress. (Currently $20/month).

I started with by asking Gemini to look up my textbook chapter and add slides to a deck

Request outline

Revise with verbatim quotes

Approve and build

Gemini foundered when I asked it to match the design to my other slides so I did it manually

Match visual design

Check accessibility

To carry out these steps, Gemini and Claude have to browse, copy and paste, carry out commands in Google Slides and PowerPoint.

Gemini again foundered when I asked it to check quotes against the original source, so I switched to Claude Cowork

Check against the source

Refine citation format

To accomplish this, Claude in Chrome navigates back to the textbook, writes code to find similar text and determine if it matches, change it to match, and insert links.

Auditing and confirming accessibility

Next, I asked Claude to use a checklist from AccessibilityUnraveled.com. I ran the PowerPoint check.

Review audit results

Pick which fixes to apply

Confirm with PowerPoint Checker

One of the resulting slides

Part of Claude Cowork’s task list as it created a spreadsheet of AI about OER

Disclosure is key. A Sample AI Use Statement

“The above template phrases followed by asterisks (*) were adapted from ChatGPT output responding to a "Template phrases for reflecting on AI feedback prompt," ChatGPT, 25 Sep. version, OpenAI, 8 Oct. 2023, https://chat.openai.com/share/9c526e9b-b654-4c24-abb3-01a173026a63. The remaining original phrases and the organizational structure are by Anna Mills and are shared under a CC BY 4.0 license.”

​

— Section 16.1: Template Phrases for Reflecting on AI Feedback — Anna Mills, How Arguments Work — A Guide to Writing and Analyzing Texts in College (Mills, 2025a)

AI and open licenses

• Open licenses help us value and acknowledge human contributions
• AI outputs themselves can't be OER; only the ways we shape AI outputs can.
• You can only license your contribution to the result, not the AI’s contribution.
• See the January 2025 Copyright and Artificial Intelligence report of the U.S. Copyright Office.
• To the right is a sample licensing statement from Creative Commons.

Attribution and licensing resources

• Artificial Intelligence (AI) Attribution and Citations for Open Educational Resources (OER) Frequently Asked Questions (FAQ)
• Understanding CC Licenses and Generative AI
• Affordable Learning Georgia’s Guidelines: Generative AI Tools in Open Educational Resources

Mentimeter Poll

• How do you use AI or hope to use AI in OER work? Or why do you choose not to use AI?
• What kinds of AI use do you disclose, and how do you disclose?

​

Use the arrow key in the Mentimeter or go to https://www.menti.com/alwd1nt5hu4d (link in the chat) or to Menti.com and enter 6509 9225

​

Questions and comments?

I hope we can continue the conversation on LinkedIn, BlueSky, and Substack!

​

Slides: https://link.annarmills.com/OpenCon

​

This slide deck is licensed CC BY NC 4.0.