I Know It's A Lot More Than Just Being Bored
In Which I Offer a Running Timeline of My AI Usage On One Day in April
As someone who (mostly) advocates for thoughtful AI engagement—and in many cases careful adoption—I figured it might be helpful to share explicitly how I'm integrating AI tools into my daily work. My role primarily involves supporting faculty, though I also teach one section of 11th grade U.S. History. Much of what follows is directly relevant to teachers’ professional lives outside the classroom, yet I’ve also recently identified what I think are clear opportunities for meaningful classroom integration.1
This reflection was prompted by a recent conversation with my head of school, and I'm lucky enough to work in an environment that actively encourages AI exploration. Given my naturally curious disposition, this freedom to investigate and experiment is a true gift.
Below, I outline some general insights about my ongoing learning in this area. However, the most illuminating part for me—and likely for you as well—is a detailed snapshot of a single day's interaction with various AI tools. Before journaling this, I honestly didn't appreciate just how embedded AI has become in my routine. Importantly this isn't really about offloading cognitive effort. Instead, (I think) it reflects ways I’m seeking to thoughtfully extend my capabilities. Indeed, in at least one instance, I outright rejected the AI’s recommendations.
If you're not currently leveraging AI in your workflow, you might find value in journaling your own routines to discover points of possibility—or not. It’s your call. It’s a free world and you have agency. I’m not suggesting any of these things will work for you. But they increasingly work for me.2
How I'm Learning about AI
Social Media & Real-Time Insights: Twitter/X remains my primary resource for real-time, cutting-edge insights into AI developments. Despite occasional frustrations with the platform, the immediacy and quality of discourse make it indispensable for staying ahead in this rapidly evolving field.
Podcasts & Thought Leaders: Dwarkesh Patel’s podcast consistently features discussions with leading AI thinkers, cutting through hype to address foundational issues. Similarly, economist Tyler Cowen—through his Twitter/X feed, podcasts, and Marginal Revolution blog—offers consistently fresh perspectives on AI, benefiting from his longstanding engagement with the topic.3
Substack & AI Community: Substack has emerged as a vibrant intellectual ecosystem for thoughtful AI analysis. I regularly follow influential voices like Ethan Mollick, Michael Lomuscio, Eric Hudson, Rod Naquin, Peter Nilsson’s Educator’s Notebook, and Marc Watkins, whose diverse and insightful commentaries enrich my understanding and spark new ideas. Teacher Steve Fitzpatrick’s Substack is new to my diet and certainly worth following too. I follow many more critical voices as well. I learn a lot in those spaces and it allows me to check my priors and to balance my own thinking.
Hands-On Experimentation: Perhaps most importantly, I actively engage with AI tools daily. Treating AI as an instrument, I frequently experiment and explore new possibilities, pushing the boundaries of my own capabilities alongside the technology. This iterative, playful approach allows me to develop practical expertise and uncover innovative uses of AI in my work. The daily journal and timeline below demonstrates how this often plays out each day.
My Day with AI (April 9th, 2025)
5:45 AM—Proofreading, Editing, and Image Creation: Conducted a final proofreading and editing session for this week's WDYN post using ChatGPT 4.5. I draft these pieces fully in my own voice, then refine them based on AI feedback. Critically, I don't always adopt suggested changes, as they sometimes dilute my authentic tone and style. I also tweaked the image we use on The Beacon using Midjourney, because the one I landed on the day prior didn’t quite capture the essence I was hoping for related to the post. This (5:45 AM) is not hero ball—I just like to look at anything like this multiple times and as late as I can before deadlines. Plus my kids aren’t typically up at this time, so I have some peace and coffee before the chaos unfolds.
8:30 AM—Virtual Note-taking: Met virtually with a Director of Technology at peer school to discuss AI policies and integration with Flint. Used Fireflies AI to capture meeting notes automatically. With numerous virtual meetings weekly, Fireflies significantly streamlines my workflow by summarizing action items and critical next steps, which I typically review each Friday afternoon. It’s always a nice snapshot of my week and I am able to review action items and identify next steps on a series of projects.
9:30 AM—Experimentation with Essay Feedback: Explored grading and feedback automation with GPT 4.5. Initially, I blind-read student essays, assigning brief notes and preliminary grades. The rubric itself had previously been generated in collaboration with GPT 4.5, based on the original assignment criteria. Next, I uploaded the student essays, prompting GPT 4.5 to generate class-wide feedback—remarkably, its analysis aligned closely with my own. Subsequently, GPT provided individual feedback and assigned grades, resulting in striking similarity: 13 students received identical grades to mine, one student's grade was higher by a third, and two received grades lower by a third. While I ultimately used my own grades, this experiment reinforced the significant potential of AI for streamlining teacher workflows and enhancing efficiency in providing timely, effective student feedback.4
10:15 AM—Lesson Planning Tweaks: Uploaded previous WWII lecture notes to GPT 4.5 for refinement and to create engaging learning activities. The AI helped craft a compelling opener involving a comparison of two World War II images, generated a thoughtful mid-lecture question examining US isolationism historically and today, and suggested an impactful closing activity prompting students to create a headline encapsulating the early war years. This streamlined process enriched the lecture’s interactive elements effectively. Nothing groundbreaking here, but given the end-of-year time crunch, this thought partner process helped me quickly reorganize and enhance lecture slides with minimal stress in what was a busy day.
11:30 AM—Conversational Computing Partner: Uploaded concise notes from my regular 1-on-1 meetings with a colleague into NotebookLM. The goal is to surface key insights and identify trends across the year, pinpointing areas of CTL growth, academic opportunities, and personal development. I used NotebookLM like a ramped up Google Docs. I’ve often lamented that Google Docs is a place where good ideas go to die. The dialogic computing aspect of NotebookLM provides an additional perspective and kind of thought partnership to reveal meaningful patterns and actionable insights. I frequently revisit folders and projects in NotebookLM in ways that I don’t in Google Documents.
2:45 PM—Microlearning Script Creation: Leveraged GPT o1-Pro to craft a microlearning script for 7taps, inspired by a colleague’s shared Aeon article, “What Actors Know: Acting is an ancient and intrinsically human way to establish vibrant connections with one another. And it’s available to us all.” o1-Pro excels at synthesizing content and optimizing it specifically for the 7taps platform, enabling efficient development of engaging, concise learning modules. We're currently in the Minimal Viable Product (MVP) phase, practicing extensively to ramp up production of content for next year.
3:30 PM—Let’s Get Meta: Utilized GPT 4.5 to assist in crafting this document, inputting loosely formed sentences and core ideas to streamline the writing process. This approach allowed me to efficiently produce clear, polished content without unnecessary stress—ensuring authenticity, while maintaining practicality for informal communications. In the end, this is probably 80% my writing and 20% AI generated text. I am being transparent with you about this. I am okay with this. You may not be. I am not here to persuade you otherwise.5
6:30 PM—Research Deep Dive: My kids are at KidStrong (think CrossFit for kids), giving me an hour of downtime. Earlier in the day I met with key colleagues about summer reading options for faculty. It was a productive discussion, though we struggled with how best to integrate this learning approach into our adult professional community. With this context in mind, I used Elicit—an AI-powered research assistant we piloted last year and now frequently use—to rapidly generate a research report addressing the question, "What is the role of book reading in adult professional learning?"
In just 12 minutes, Elicit scanned 500 research papers, identified and analyzed the top 25, extracted over 150 relevant data points, and created a detailed, multi page report of the findings. This process would typically take hours or even days manually. Notably, I can also interactively "chat" with the generated report, exploring key insights and identifying further areas of interest. I'll review the full report later tonight or the next day, but the preliminary summary was illuminating: "Book reading, when combined with regular discussion and relevant texts, serves as an effective tool for adult professional learning by fostering communities and facilitating knowledge transfer."
This experience also highlights a broader truth: AI isn’t any one thing. Rather, it's a diverse array of specialized tools, each uniquely suited to empowering specific workflows. The day’s journal captures just how varied and impactful these tools can be. Though some might describe me as “advanced” in terms of AI usage, today represented approximately a "60% AI" day. I utilized five distinct AI tools—Midjourney, Fireflies, ChatGPT, NotebookLM, and Elicit—and employed multiple models within ChatGPT alone. This level of integration is the product of deliberate learning, constant reading, and active imagination about how AI might enhance my work.
That's my current AI stack—not a prescription, just transparency about what works for me. Ultimately, these tools haven't just tweaked my workflow; they've fundamentally reshaped its landscape. I am certainly not bored.
Note: I always pull blog titles from song titles or lyrics. It’s a thing. It’s fine. Just go with it. I named the blog ‘The Academic DJ’ after all. Today’s title comes from Titus Andronicus’s ‘Ecce Homo.’ I like the lyric “I know it’s a lot more than just being bored,” precisely because I don’t explore this stuff because I am bored. I do so because I think it matters and I derive lots of learning from it. The song also rips and I always listen to a lot of Titus Andronicus in the spring time for some reason.
I recently wrote about that here.
And I really mean “work for me.” None of the examples I provide are fully agentic. Elicit is close and the one tool I would really miss if I no longer had access to it. But when I leverage these I am doing so in a way that positions me as a conductor rather than a spectator—I orchestrate the inputs, guide the process, and critically evaluate the outputs to enhance my own thinking, productivity, and decision-making.
Some of you might already be bristling at my consumption of what many describe as “tech bro” and Right/Libertarian leaning sources. That’s feedback for you. Not me.
Note: I turned off data capture when doing this and I removed all identifying student information.
🤷🏻