Automate the Small Stuff
Automate the Small Stuff
Justin Grandinetti
Two years since the launch of ChatGPT as AI’s “killer app,” opinions about the integration of generative AI into college and university curricula has largely bifurcated.
There are some who embrace generative AI as an inevitable and essential skill, whereas others engage in principled resistance, often due to a mix of concerns about academic integrity, student rigor, and the erosion of essential skills. Taken too far in either direction, these diametric positions run into limitations; there is as much at risk in uncritically mimicking big tech hype cycles about the limitless disruptive power of generative AI as there is in hoping that the technology is simply a passing craze. We can only speculate about the lasting impacts of large language models, but for the time being educators are left with a critical question: where should we draw the line on acceptable generative AI use in the classroom?
Perhaps an answer is to encourage students and faculty to automate the small stuff.
One strategy I’ve used to introduce students to the productive use of generative AI is through a concision assignment that blends critical reading, writing skills, and AI-assisted editing. Students begin by selecting one of three assigned academic articles, each ranging from 6,000 to 8,000 words. Their first task is to read the article closely, take notes, and manually condense its key ideas into a 1,000-word summary. Their next step is to experiment with how generative AI can attempt the same task by trailing different prompts, documenting their process, and analyzing how the model interprets and distills complex arguments. Students then take the AI-generated summary and refine it themselves, identifying errors, omissions, and stylistic weaknesses. Finally, students engage in reflection: What editorial decisions did they make? How effectively did the AI handle a similar task? How did adjusting their prompts shape the AI’s responses? And ultimately, what human intervention was still necessary to produce a high-quality summary? Through this exercise, students engage critically with how they can work with generative AI—not as a replacement for their thinking, but as a means of enhancing their editorial judgment.
In this concision assignment, what matters most is not just the final product, but the steps students take to get there.
The goal of this assignment for students is not merely in the various edited outputs, but in the process and contemplation. Before they can meaningfully summarize an article—whether manually or with AI—they must first engage deeply with the material, determining its core arguments and significance. AI might assist in streamlining language and structure, but it does not replace the cognitive work of comprehension. In this model, what gets outsourced is some labor, not thought. The student remains the editor, the evaluator, and the final decision-maker.
The same logic applies beyond the classroom.
As university educators, we frequently take on administrative tasks that, while necessary, are often repetitive and time-consuming. Writing responses to evaluations, summarizing reports, and drafting institutional memos require thought and consideration, but they also follow familiar, structured formats. In such cases, using generative AI to produce an initial draft of certain sections is not to relinquish intellectual work. It is instead automation of routine so that we can focus on refining, clarifying, and ensuring the material aligns with our intent.
My concision assignment offers a constructive vision for generative AI integration—not as a replacement for human thinking but to free up time for deeper intellectual engagement. After all, we have for some time accepted technology like sentence autocomplete in Gmail and the built-in synonym feature in Microsoft Word. We’ve all already become cyborgs, seamlessly integrating digital technology into our workflows. At its best, AI can take over the small, time-consuming tasks, provided we set the parameters and remain actively involved in shaping the output. As proof of concept, this very editorial was co-written with generative AI. I structured my thoughts in bullet points, outlining the sentence and paragraph progression much as I would for any academic article. The AI assisted in shaping the prose, but the final product required some substantial editing—removing deterministic language, cutting unnecessary platitudes, ensuring clarity and nuance, and even scrapping and re-writing some initial ideas that didn’t make the cut. While I am far from trusting AI to generate the type of precise writing required for a quality peer-reviewed journal article, I see immense value in hybridizing writing production for everyday tasks. By approaching generative AI with intention, we preserve cognition, not lose it. Or, put another way: we automate the small stuff.