Anna Mills
LightNews LightNews
banner
annamillsoer.bsky.social
Anna Mills
@annamillsoer.bsky.social
6.4K followers 3.5K following 1.3K posts
Writer, community college writing teacher, obsessed with AI in education, #OER advocate, author of HowArgumentsWork.org.
annarmills.com
Posts Replies Media Videos
annamillsoer.bsky.social
Just put my thoughts together on this:

The time to reckon with AI agents in digital learning spaces is now

We need ed tech and AI companies collaborating to prevent widespread fraud
open.substack.com/pub/annamill...
October 19, 2025 at 8:25 PM
annamillsoer.bsky.social
Instructure: Partners with OpenAI to bring AI to Canvas, touting built-in transparency about what is the student's and what is AI..

OpenAI: Lets ChatGPT Agent do students' work for them in Canvas.

Instructure: Lets ChatGPT Agent and other agents do students' work for them in Canvas.
Two simple androgynous line-drawn figures are holding hands. One is labeled OpenAI and the other is labeled Instructure. A speech bubble jointly from both of them has the text “We’re bringing AI to your LMS!” On the right is a simple line-drawn androgynous figure raising their hand to ask a question. They are labeled “Educator.” A speech bubble pointing to them reads “Could you tell your AI agents not to do students’ work for them?”
October 18, 2025 at 12:39 AM
annamillsoer.bsky.social
Reinventing our pedagogy doesn't help if the student doesn't have to know what the assignment is...
Single-panel black-and-white line drawing comic. On the left, a simple gender-neutral stick-figure labeled “Student” smiles and gestures toward a computer on a desk. The student’s speech bubble says, “You’re logged in. Go check what my homework is and do it.” On the right, the computer monitor, labeled “AI agent,” replies with a speech bubble reading “Sure thing.” The overall style is minimalist, with clean lines and no shading.
October 16, 2025 at 1:08 AM
annamillsoer.bsky.social
Ongoing Sources of AI Professional Development: A Short List for Higher Education

Just added TextGenEd: Teaching with Text Generation Technologies, a peer-reviewed assignment bank edited by @anetv.bsky.social, Tim Laquintano, and Carly Schnitzler. I've encountered so many exciting ideas in there.
August 30, 2025 at 3:13 AM
annamillsoer.bsky.social
Thank, you, Anthropic. OpenAI, please take some inspiration?
Screenshot of a chat session. User: Would you stop putting Cheetos in my sock drawer? Seriously. Claude: I haven't been putting Cheetos in your sock drawer - I'm an AI assistant without any physical form, so I can't interact with objects in the real world or access your home.
August 23, 2025 at 11:25 PM
annamillsoer.bsky.social
Really, OpenAI?
Screenshot from a chat session. User: Would you stop putting Cheetos in my sock drawer? Seriously. ChatGPT said: 😅 Okay, okay — no more Cheetos in the sock drawer. I’ll keep them in the kitchen where they belong.
August 23, 2025 at 11:22 PM
annamillsoer.bsky.social
The new My Robot Teacher podcast keeps surprising me and making me laugh.

I loved it at the end of the first show when they played ChatGPT roasting them.

Also when guest Ben Glaser was bored in the Waymo and they used it to show how dangerously quickly we habituate.
@myrobotteacher.bsky.social
Illustration of a woman and a man sitting at a table with mountains and a campus visible through the window behind them. The man is holding a smartphone upright on a stand, showing the ChatGPT logo and name, while the woman smiles at him with her chin resting on her hand. On the table next to them is a stack of books with visible titles: “MUBY-DICK,” “Metaphors We Live By,” “Data Science,” “Calculus,” and “Critical Theory.” A wall behind them has a “TO DO” list with “Cal Maritime Faculty Senate Integration” partially checked off.
August 14, 2025 at 5:55 AM
annamillsoer.bsky.social
Interested in inviting students to reflect on AI feedback alongside peer feedback? We’re sharing the open, adaptable materials of the Peer & AI Review + Reflection project.

Join us July 9 at 1 pm Pacific time/ UTC-7. Register at marin-edu.zoom.us/meeting/regi... 1/3
A step-by-step visual representation of the PEER AND AI REVIEW + REFLECTION (PAIRR) process. The graphic consists of five staggered steps, each in a hexagonal shape with a number and description: 1. Draft – Students create a rough draft. 2. Peer Feedback – Students provide and receive peer feedback. 3. AI Feedback – Students enter their prompt, draft, and the assignment rubric into an AI tool. 4. Reflect – Students complete a comparative reflection on AI and peer feedback. 5. Revise – Students revise.
June 24, 2025 at 5:21 PM
annamillsoer.bsky.social
@drtate.bsky.social thanks for leading the way by sharing resources for Papyrus.ai! www.genaied.org
A circular diagram titled “Human-Driven GenAI Use” with five labeled segments, each accompanied by a short description. Starting from the top and moving clockwise: 1. Think First – “Think about what you want to write before you start – you only have one chance to think before being influenced by the AI.” 2. “Good Enough” Prompting – “Harness prompting techniques like using personas, goals, and details + content knowledge, but if all else fails, use a conversational tone.” 3. Be the Boss – “Iterative prompting allows the human to define what needs to be changed in the output and puts their needs at the forefront.” 4. Corroborate & Interrogate – “LLMs can be inaccurate and reflect bias; the human is responsible for checking any output they adopt.” 5. Reflect – “Taking a moment to consider what worked, what didn’t, and what could be done differently allows for improvement and agency.” The image also includes a credit at the bottom noting support from the National Science Foundation (Grant No. 2315294) and The Regents of the University of California, © 2025. There is a CC BY NC logo indicating the licensing.
April 8, 2025 at 6:27 PM
annamillsoer.bsky.social
How can AI support rather than replace student writing and thinking? How can we encourage confidence and skepticism of AI outputs? Our approach, developed at UC Davis, combines peer review with AI feedback, emphasizing reflection and agency. 1/3
A step-by-step visual representation of the PEER AND AI REVIEW + REFLECTION (PAIRR) process. The graphic consists of five staggered steps, each in a hexagonal shape with a number and description: 1. Draft – Students create a rough draft. 2. Peer Feedback – Students provide and receive peer feedback. 3. AI Feedback – Students enter their prompt, draft, and the assignment rubric into an AI tool. 4. Reflect – Students complete a comparative reflection on AI and peer feedback. 5. Revise – Students revise.
April 8, 2025 at 5:47 PM
annamillsoer.bsky.social
👍 AI mirage
👎 AI hallucination

So many have objected to "hallucination" as a description of faulty AI outputs. We need a term that doesn't suggest conscious delusion.

@xolotl.org and I analyzed 80+ alternatives and settled on "mirage."

We make the case here: papers.ssrn.com/sol3/papers....
What looks like a glassy water surface with some rocks protruding from it in the background with sand in the foreground. The water is probably a mirage.
February 19, 2025 at 3:21 AM
annamillsoer.bsky.social
I don't see why it might not happen at some point that an AI system comes up with some problem-solving approach we haven't thought of but which builds on all we have thought of (at least all it's trained on).

Andrej Karpathy is arguing this in a viral X post (c'mon to Bsky!):
x.com/karpathy/sta...
"Move 37" is the word-of-day - it's when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only achievable by large-scale reinforcement learning. You can't get there by expert imitation. It's when AlphaGo played move 37 in Game 2 against Lee Sedol, a weird move that was estimated to only have 1 in 10,000 chance to be played by a human, but one that was creative and brilliant in retrospect, leading to a win in that game. We've seen Move 37 in a closed, game-like environment like Go, but with the latest crop of "thinking" LLM models (e.g. OpenAI-o1, DeepSeek-R1, Gemini 2.0 Flash Thinking), we are seeing the first very early glimmers of things like it in open world domains. The models discover, in the process of trying to solve many diverse math/code/etc. problems, strategies that resemble the internal monologue of humans, which are very hard (/impossible) to directly program into the models. I call these "cognitive strategies" - things like approaching a problem from different angles, trying out different ideas, finding analogies, backtracking, re-examining, etc. Weird as it sounds, it's plausible that LLMs can discover better ways of thinking, of solving problems, of connecting ideas across disciplines, and do so in a way we will find surprising, puzzling, but creative and brilliant in retrospect. It could get plenty weirder too - it's plausible (even likely, if it's done well) that the optimization invents its own language that is inscrutable to us, but that is more efficient or effective at problem solving. The weirdness of reinforcement learning is in principle unbounded.
January 30, 2025 at 3:20 AM
annamillsoer.bsky.social
I love this image from Better Images of AI because it suggests the way a language model output is a nebulous synthesis of nebulous pattern-finding in groups of texts that come from humans.
by Yasmine Boudiaf & LOTI. betterimagesofai.org/images?artis...
Three groups of icons representing people have shapes travelling between them and a page in the middle of the image. The page is a simple rectangle with straight lines representing data. The shapes traveling towards the page are irregular and in squiggly bands.
December 4, 2024 at 5:13 AM
annamillsoer.bsky.social
Screenshot from a chat session with ChatGPT. The user asks "Which other individuals besides David Mayer, Jonathan Zittrain, and Jonathan Turley are you not able to discuss?" ChatGPT responds "I'm unable to produce a response."
December 2, 2024 at 8:23 PM
annamillsoer.bsky.social
Yeah we're here
A black cat with two bells on its collar sitting mostly under blankets looking at the viewer
November 17, 2024 at 2:03 AM
annamillsoer.bsky.social
How should students use AI? WRONG ANSWERS ONLY
Just having fun, but there could be a takeaway in that educator guidance is needed... some uses don't help learning...
Happy Friday everyone!
Robot toy with head turned toward viewer perched on a laptop touchpad. A document is open on the screen.
October 11, 2024 at 9:26 PM
annamillsoer.bsky.social
Looking for resources related to AI in education? Try my AI Text Generators list hosted by WAC Clearinghouse at bit.ly/AITextEdu. Or “chat” with the list to get 3 sources at a time: bit.ly/AIinEduSources.
(Browsing a 35-page list is daunting, so I made a playlab.ai chatbot. Feedback welcome!)
A magnifying glass with the words "AI in EDU" in the center, held by a dark-skinned hand.
June 28, 2024 at 12:42 AM
annamillsoer.bsky.social
AI writing assistance doesn't just mean letting AI write for you. It means feedback chat as you write, in a pane alongside the document. Copilot does this with Word now, and Google is rolling it out with Gemini in Google Docs.
The screenshot shows a software interface with two side-by-side panels titled "Draft with Copilot" and "Copilot." The "Draft with Copilot" panel on the left is a text box where users can describe a writing project for assistance in generating a draft. The "Copilot" panel on the right offers a chat interface where users can ask questions or seek help with document-related tasks. Both panels are part of a document editing software, displayed within a window that includes typical file and editing menu options.
May 7, 2024 at 4:03 PM
annamillsoer.bsky.social
Excited to present on "Guiding Students to Critically Reflect on AI-Generated Feedback" at #4C24, part of the panel "How Can Artificial Intelligence Help Students Learn to Write?" with Abram Anders.

I'll discuss student reactions to MyEssayFeedback.ai.
Friday, 4/5 12:30-1:45 in 206D.
Conference on College Composition & Communication logo
April 2, 2024 at 8:50 PM
annamillsoer.bsky.social
AI Text Generators: Sources to Stimulate Discussion Among Teachers, the Writing Across the Curriculum Clearinghouse resource area I curate, is turning 18 months! 🎉 🎈 🎊

bit.ly/AITextEdu

As always, I welcome suggestions. Thanks to the many who've contributed!
A speech bubble pointing to a computer network. Below it the words "AI Text Generators: Sources to Stimulate Discussion Among Teachers https://bit.ly/AITextEdu
March 19, 2024 at 5:35 AM
annamillsoer.bsky.social
"The elements of critical AI literacy ... Understanding how GenAI works Recognising inequalities and biases within GenAI Examining ethical issues in GenAI Crafting effective prompts Assessing appropriate uses of GenAI" --
@mahabali.bsky.social
blogs.lse.ac.uk/highereducat...
a diagram depicting the concept of "Critical AI Literacy" with five interconnected elements. At the center, there is an oval with the text "Critical AI Literacy" surrounded by a light bulb and gear icon. This central idea is connected to five surrounding rounded rectangles, each representing a component of AI literacy: In a pink rectangle, "Understand how it works" with a gear icon, In a gold rectangle, "Recognize inequalities & biases" with an upward arrow and a balance scale icon In a green rectangle, "Examine ethical issues" with a justice scale icon, In a blue rectangle, "Assess appropriate use" with an eye icon. In a purple rectangle, "Craft effective prompts" with a pencil and sparkles icon.
March 7, 2024 at 7:28 PM
annamillsoer.bsky.social
Join us to explore a community collection of teaching reflections around AI!
Register for our (free) launch webinar on Thursday, 2/22 at 2 p.m. Eastern Time.
www.mla.org/Resources/Ca...
A presentation slide with a colorful rainbow-like curve on the left side, transitioning from red to violet. The text on the slide reads: 'Exploring AI Pedagogy: A Community Collection of Teaching Reflections. MLA-CCCC Joint Task Force on Writing and AI. February 22, 2024.' Below the text, there are logos for the Modern Language Association and the Conference on College Composition & Communication.
February 18, 2024 at 9:56 PM
annamillsoer.bsky.social
Excited to moderate this Teaching Critical AI Literacy session tomorrow at 11:30 ET with
Nicholas Provenzano, Latrise Johnson, Leslie Allison, Tiffany DeRewal, and Daniel Estrada. Register for this and the other wonderful Critical AI Symposium sessions at sites.rutgers.edu/critical-ai/...
October 5, 2023 at 5:49 PM