An Ounce of Prevention
How I'm engaging generative AI in my teaching this fall
Thank you for being here. As always, these essays are free and publicly available without a paywall. If you can, please consider supporting my writing by becoming a patron via a paid subscription.
As the spring semester begins to fade and the summer rhythms take root I'm beginning to think about the next academic year and what I'm planning to change in my courses. The big question for many is what to do about AI.
Last summer around this time GPT-4 was out, but still behind a $20/month paywall with limited access. Anthropic's Claude was still on version 2. Google's forefront model was Bard. Things are different now. Today we have several capable models and they are all much more accessible than in the recent past, often available without any cost to the user at all.
Not only are the latest generative AI tools more capable and accessible than ever but it's become clear that educational spaces are going to be getting a lot more attention moving forward. Both OpenAI and Google have not so subtly hinted that they're gunning for education.
All this means that it's more important than ever for educators to get their heads around generative AI and how it impacts their work before the fall. Today I want to give you a peek into my headspace about generative AI and the ways that I’m planning to engage it, not embrace it this fall. What may surprise you is that my plan has little to do genAI itself and much more to do with redoubling my commitment to the art of teaching.
The Absent-Minded Professor is a reader-supported guide to human flourishing in a technology-saturated world. The best way to support my work is by becoming a paid subscriber and sharing it with others.
At a minimum, every teacher from high school on up should include a generative AI policy as part of their syllabus this fall. If you’re curious, you can read the one I drafted last year here. But beyond syllabi policies, it also is worth seriously considering how you might engage with generative AI in your discipline. I intentionally say engage and not integrate. These tools are significant enough that we all must engage with them enough to understand how they work and to address their impact on teaching and learning. But before we move to integration, we had better consider how and why we want that integration to happen.
As much as the crush of new genAI tools is going to require us to adapt, I think there's a silver lining: this disruption offers us an opportunity to revisit and renew our core values about the teaching and learning that is going on in our classrooms.
Today I want to share a few thoughts about how I'm approaching genAI in my context with the hope that you might find something that will translate to yours. Here they are:
Build trust and a collaborative learning environment
Embrace transparent teaching
Cultivate intrinsic motivation
Revisit assessment practices
Thoughtfully experiment with genAI
As always, please leave a comment or send me an email if anything resonates or if you have a follow-up question.
1. Build trust and a collaborative learning environment
Building trusting and collaborative relationships with students is one of my core values as an educator. I want every student in my class to understand that I am here to know them as a person and to help them to engage in productive struggle with the material in the course.
The first step to building trust is to encourage candor about each person's objectives in the course. Why are you here and what are you hoping to get out of it? Every student will come to classes with a different set of priorities and goals: everything from "I'm required to be here" to "I've been waiting to take this class for the last four years." Of course, we want to get all of our students to a place where they are invested in the course and learning the material, but the only way to get there is to start where students are.
The first step toward engaging each student is laying the foundation for a relationship. In addition to sharing your hopes for the class, ask them to reflect on what they want out of it. Even if they're only there because it's a required class, making that clear is important.
What I’m doing
I've tried this before in various ways with surveys, but this fall I'm thinking of taking it one step further. I'd like to prototype a learning journal where students spend some time in the first week of the semester jotting down their goals for the class and then regularly revisiting them to see how they're doing.
2. Embrace transparent teaching
Shortly after ChatGPT was released to the public in the fall of 2022 I wrote a piece about how transparent teaching plays a big part in my approach to the classroom and how I felt it was going to be especially important in a post-ChatGPT world. A year and a half later, I remain convinced that transparent teaching is a valuable approach that addresses some of the challenges posed by genAI head-on.
The main goal of transparent teaching is simple: to promote students’ conscious understanding of how they learn. This framing means that as educators we need to be clear not only about what we hope our students are learning but also about how and why.
As the name suggests, transparent teaching is about making things clear to students. Many times the motivation behind a particular assignment is unclear to students. One way to address this is to communicate the learning objectives for every assignment.
But the part of transparent teaching that is even more valuable in a world where ChatGPT is accessible with a quick keyboard shortcut is the focus on communicating the how and why of what's going on in the classroom. It may be that some of the assignments in your class need to change in response to ChatGPT. But if they need to change, it's likely that they already needed to change before ChatGPT. For many other assignments, it's possible that they don't need a full overhaul but rather a re-justification to give students more clarity around the purpose of the assignment and the reason that it is a worthwhile use of a student's time.
At this point, you might see a connection between transparent teaching and building relationships of trust. These first two points reinforce one another: strong teacher-student relationships build relational capital to encourage students to try assignments even if they might not completely buy into the methodology at first and transparent teaching helps to enable students to understand and offer feedback on questions or disagreements they might have about particular assignments.
What I’m doing
A few years ago I made a concerted effort to go through all my assignments in my embedded systems class to add explicit learning goals. I plan to revisit those this summer. I also want to add some material at the beginning of class to more clearly communicate the rationale behind why the course is structured the way it is and to highlight how the structure embraces good pedagogical practices.
3. Cultivate intrinsic motivation
I'm a strong believer that intrinsic, rather than extrinsic motivation should be the goal. I'm not arguing that extrinsic motivation is intrinsically bad (please forgive me for the wordplay), but I do think that extrinsic motivation is much more fragile than one rooted in a love for the course content and a desire to learn it for its sake rather than to earn a grade or to achieve a certain outcome.
This is one area where a personal approach to education is particularly important. Just as each student comes to a class with a unique set of goals, they will also pursue these goals for different reasons. Some students are there because they enjoy the content of the course and want to dig deeply into it. Others might be there as part of their larger goal to get a degree to get a well-paying job to support their family or their future ambitions. Others may be there because their parents told them to be there. Each of these sources of motivation comes with different core values underneath them and also with a different ability to encourage students in their learning journeys.
Again, this point is connected to the others: cultivating intrinsic motivation needs to start with building trust with students, demonstrating transparency in our communication with them, and asking for candor from them. Whatever students share about the source of their motivation is valuable, as long as it is honest. It does neither the teacher nor the student any favors to pretend. Of course, students may not feel comfortable sharing their motivations, especially if they are only there because the class is required, but starting by making the implicit explicit, at least for the student, is valuable.
What I’m doing
My hopes for this fall are small steps in this direction: I see an investigation of sources of motivation as playing nicely into a self-assessment performed by each student early in the semester. Perhaps I will collect these anonymously as well.
4. Revisit assessment practices
Of all the areas of teaching, assessments might be the most significantly impacted by genAI. But genAI might be doing us a favor here. If genAI disrupts an assessment, it may not be a good one in the first place.
Rethinking genAI's impact on assessment is way too big to tackle in a few short paragraphs, but my fundamental question here is again to revisit the learning goals. What is the point of any given assignment and how are its goals impacted by genAI? If you are teaching a writing class, genAI tools may pose a more significant threat to your preexisting assignments, but only if students are already predisposed to see the process of writing as a means to the end of the product. If your assessments are already aligned to reward the process (e.g., allowing multiple revisions without penalty), then the potential temptation to use tools like ChatGPT may be lessened.
What I’m doing
In my work, I've been experimenting with alternative grading strategies and implemented a specifications grading scheme in my embedded systems class last fall. It will likely not surprise you to hear that this too is connected to the other points I shared above: I found that specifications grading gave me new ways to build trust with students, clearly communicate the goals for my assignments, and help develop students' intrinsic motivation.
If you are interested in learning more about alternative grading, I highly recommend the
newsletter (and book). If you're looking for a place to start, you can read my guest post reflecting on my first experiment with alternative grading last fall.5. Thoughtfully experiment with genAI
My final point is a call to action: we need educators to be a central part of the conversation around how genAI gets implemented in education. We are at the point of contact with students and those interactions offer significant opportunities to understand the impact these new tools will have in the classroom.
If you're considering experimenting with genAI in your classroom this fall, I would encourage you to do it. As you experiment, think about how you can invite your students into the process of exploring these tools along with you. You may have some rough ideas about how to incorporate these tools into your course. As the instructor, you also bring an expert perspective on the overall course learning objectives and pedagogical practices at play. With that said, don't discount the role your students can play in exploring with you. If you have built a foundation of trust with them with clear and transparent communication of the course learning goals and methods, consider giving your students some freedom to experiment themselves.
The approach I take in my classes is to encourage students to explore these tools freely with the caveat that they need to do so within a structure I provide. The general process is for them to articulate what they want to use genAI for and why, to document their actual use of the tool (especially if it shifts from their original intent), and to reflect on their experiment. One way to capture these is through 4 Cs:
Clarify: What is the problem you're trying to solve?
Contemplate: Why do you think AI will help?
Catalog: How did it go? What did you do? How did your approach shift?
Communicate: What did you learn? How will you share?
Coming to a classroom near you
genAI is here. It's almost guaranteed to have touched your students already, even if you aren't aware of it. While genAI presents some threats, it also creates opportunities. This fall I'm using genAI as an opportunity to double down on my core values as an educator: building relationships of trust with students to help them to explore the fullness of who they are and who they can become.
Reading Recommendations
writing at is a great source for thoughtful commentary on genAI and how it might impact your class. One of his recent posts on the impact of automating feedback is no exception.AI feedback can give a user some thoughtful, even insightful suggestions about where to take a piece of writing, but its feedback is limited to the generic and aggregate average of what AI predicts a user would want to hear. There is no personality, no intent, no ability to shape or persuade an author into changing their writing.
This week Fei-Fei Li and John Etchemendy (both at Stanford’s Institute for Human-Centered Artificial Intelligence) wrote a piece about why today’s AI is not sentient. Worthwhile reading.
wrote a piece for the Boston Globe about the fundamental conflict between genAI and the purpose of education.When humans experience hunger, they are sensing a collection of physiological states—low blood sugar, empty grumbling stomach, and so forth—that an LLM simply doesn’t have, any more than it has a mouth to put food in and a stomach to digest it. The idea that we should take it at its word when it says it is hungry is like saying we should take it at its word if it says it’s speaking to us from the dark side of the moon. We know it’s not, and the LLM’s assertion to the contrary does not change that fact.
Embracing AI in the classroom is not a simple decision. There are tasks a chatbot can do pretty well that we still want our students to do themselves, for good reasons. For example, the fact that ChatGPT can summarize and analyze an article that I feed it does not mean that I no longer want students in my writing course to read articles or analyze what they read. I don’t assign summary and analysis because I need more summaries or analyses; I assign these projects because I want to help my students think through complex ideas and grapple with them. And I don’t ask my students to write papers because the world needs more student papers; I assign papers because I want my students to go through the process of figuring out what they think. The friction is the point.
The Book Nook
Finally getting close to finishing Klara and the Sun. It is another reminder of the power of fiction to shape our imagination and to help us to see the future. It touches on many deep themes about what it means to be human and what a world with increasingly sophisticated human-like machines might look like.
The Professor Is In
Great to have the majority of my summer students into the swing of things by the end of last week. Last week I had my initial one-on-one meetings with each student to discuss their goals for the summer. I continued a template I created last summer based on a version I found for Postdocs at Stanford with three sections for discussion:
Self-assessment:
Career Goals
Action Plan
If you’re interested, feel free to check out and download a copy of the form (PDF, DOCX).
Leisure Line
With the day off for Memorial Day on Monday I decided to make my favorite cinnamon coffee cake. Highly recommended. Recipe from Sally’s Baking Addiction here.
Still Life
A slightly angry looking turtle from the pond at Caltech over the weekend.
Josh, Well put. Since we read so many of the same writers, including Ishiguro and McLuhan, not surprising that I find myself nodding along. Of your five, the one I feel the least on top of is assessment. I'm all for alternative grading and emphasizing process over product, but I'm trying to work through the question of how augmented writing, where generative AI is incorporated freely into the process, changes my class's peer-review and instructor-review of drafts. No clear answers, other than to include reflection on the use of AI writing tools in the process. I look forward to reading more about your prep for the fall.
Thanks for the shoutout, Josh! And for the smart advice.