What Agency Means in the Era of Automation
We’ve reached the end of 2025! Thank you all for the kindness, support, and generosity you’ve shown me this year. In the past year, I saw the number of subscribers double for this newsletter as thousands of you read and shared my musings about AI and education. For that, I am truly humbled.
Before launching into a year-end list that has filled many an inbox, let me begin with one final thought as we prepare for 2026.
There Is No Keeping Up With AI
Andrej Karpathy, one of the founders of OpenAI, cannot keep pace with developments related to AI and coding within his field. Neither can I for education. I don’t think anyone can. Just a few months ago, Karpathy was publicly telling people he believed current agentic models weren’t capable or “there yet,” and has now shifted to feeling hopelessly behind. Claude’s Opus 4.5 changed his perspective rapidly. If even the architects of these systems feel adrift, our goal for students cannot be mere technical fluency; it must be the cultivation of human agency.
Teaching Agency and Resilience
The skills students and educators need right now aren’t AI-related in nature and cannot easily be taught through certification or so-called upskilling. That’s because they’re human. The chorus calling for education to remake itself in this age of automation misses a fundamental truth about what AI represents to learning, namely redefining tasks isn’t the answer. It doesn’t matter how fun, exciting, or engaging you make a curriculum, or how challenging you make an assessment; machine intelligence can now automate it. A student can task an AI to save them time and offload the practical struggle necessary to learn, so we need to talk about purpose, not product. We need to radically depart from learning as a series of tasks into a value-driven model that helps students form complete and messy pictures of themselves and allows them to navigate complicated systems.
Technology cannot offer students a reason or meaning to strive toward. Human beings are not a series of tasks bundled together leading to a deliverable. When we hear the term agency, we associate it with free will, but what does free will mean when algorithms and complex systems increasingly dictate our lives? The current discourse about agency defines the idea of will as the capacity to make meaningful choices and act on them in systems that increasingly try to decide for you. Imparting that to students means demystifying complex digital environments and social systems by equipping them with a frame of mind to help them navigate it all.
Create a Culture of Awareness
AI, social media, and countless digital distractions often make navigating traditional systems confusing. We can make students aware that this is often by design. Powerful corporations are in constant battle for our attention. Some of the most productive behaviors high agency people demonstrate are choosing to seek uncomfortable and unfamiliar opportunities with real individuals by establishing and cultivating meaningful relationships. As Cate Hall writes in How to Be More Agentic:
The idea is that making changes in your life, especially when learning new skill sets, requires you to cross a moat of low status, a period of time where you are actually bad at the thing or fail to know things that are obvious to other people.
It’s called a moat both because you can’t just leap to the other side and because it gives anyone who can cross it a real advantage. It’s possible to cross the moat quietly, by not asking questions and not collaborating, but those tradeoffs really nerf learning. “Learn by doing” is standard advice, but you can’t do that unless you splash around in the moat for a bit.
Teach for Growth, not Concepts
For students (and many faculty), teaching and learning is entirely transactional and thus easily offloaded. Most of the Edtech industry and AI have likewise focused on releasing automations that easily allow both students and teachers to make learning as efficient and as frictionless as a mobile payment. We may not be far off from one-swipe credentials. Modern education rarely emphasizes knowing oneself, mapping limitations, and ensuring productive struggle leads toward personal growth. Perhaps the most interesting and engaging curriculum for students to tackle is themselves. What are their goals for the moment? What can they offer to their classmates? What connections can they form that are meaningful to them now and moving forward? Henrik Karlsson’s Thoughts on Agency present a more personal take:
Agency, more broadly understood, is about having (or requires) an accurate map of your interiority and a map of the external world, so you can navigate to situations where they overlap in generative ways.
Normalize Struggle
In an era of frictionless AI, we must remind students that the friction of learning is where the growth actually happens. To that end, consider ways you can cultivate uncertainty and teach students to embrace ambiguity. Failure is often the outcome of many attempts, so work to remove the stigma around both failure and success and re-contextualize that both are crucial components to learning. At Harvard, Manja Klemenčič’s Advice to students: Enact your agency, build resilience, aptly makes this case:
Choose courses that will challenge you, even unsettle you. Don’t accept being coddled. When you choose to engage in debates, please have the intellectual curiosity to explore the topic in depth, have the intellectual honesty to recognize the merits of arguments of the opposing side, admit to the weaknesses in your own viewpoint, and have the intellectual humility to admit when you don’t know and wish to learn more.
Education is About Agency
Many students have told me they feel like college is an extension of high school, where learning happened to them. If AI indeed requires us to rethink education and what it means to learn now that machine intelligence is ubiquitous, then we need to invite students into that conversation and emphasize that college isn’t where you go to receive knowledge or a degree, but to actively engage in creating an experience unique and worthwhile to you. That can only happen if we help students identify and actualize goals that extend beyond our subject matter expertise. That’s my hope for 2026.
My Perpetual Thanks
As we end the year, I once again extend my thanks for your time, your energy, and your focus. We get so few moments where we get to think deeply about our world, and I am immensely grateful to all who’ve subscribed and read my words. I’m committed to keeping this newsletter free, but it’s hard to overstate how important paid subscribers have been to support this work. If you’ve valued what I’ve written and wish to support the future of this work, then I hope you will consider subscribing to a paid membership.
If you are interested in finding out more about the AI training I offer to higher education, K-12, speaking engagements, or just want to say hello, please drop me a line: mwatkins03@gmail.com or visit my personal website.
Most Read Posts
I spent much of this past year focused on how the AI industry has saturated students and faculty with free or low-cost tools. Whether it was new AI grading apps or AI-enabled browsers that could impersonate a student and allow them to automate an entire course, one area I think we need to continue to focus on is how this technology is challenging relationships in our classrooms, as AI continues to break traditional signals and erodes social trust.
Writing Elsewhere
I continued an advice column for the Chronicle focused on practical strategies for confronting AI in education. It’s been wonderful to connect with a broader audience outside of Substack and social media.
I also published two special issues of Thresholds in Education that I co-edited with Stephen Monroe about generative AI and education. A third is on its way soon!
Generative AI’s Impact on Education: Writing, Collaboration, & Critical Assessment
Building AI Literacy: Critical Approaches & Pedagogical Applications Surfing in a Tsunami
Finally, I joined Annette Vee and Derek Bruff to coauthor The Norton Guide to AI Aware Teaching, which should be available later in 2026. What a year it has been!
I have a start on the first column I'll be publishing at Inside Higher Ed after the holiday and it's titled "The Age of Agency," because I've long thought that's what we should be teaching anyway, and as you illustrate here, it's even more important with the presence of this technology.
It is somewhat comforting to note that my pedagogy has always aimed for these ideals, but now I’m getting more intentional about it. I’m trying not to feel resentful that there is so little support for it, but I’m going to try to focus on being part of the solution and helping colleagues to center agency .