AI Killed the Homework Era
Commentary: The damage may reach well beyond homework, putting the value of grading itself in doubt.
Homework used to give educators a window into what students understood. Now, for many online classes, it can be hard to tell whether an assignment reflects a student’s thinking at all. That shift matters more than the familiar debate over cheating: if teachers can’t trust the work they’re grading, feedback loses much of its purpose.
I saw that change as a professor of practice at Arizona State University’s Walter Cronkite School of Journalism and Mass Communication. After ChatGPT arrived in late 2022, major technology companies followed with increasingly capable chatbots and, later, AI agents. In my advanced communications courses, discussion posts began drawing on elaborate references beyond the assigned reading. Some students seemed to arrive as instant experts, then struggled when asked to explain how they had reached their conclusions.
The practical problem was relentless. I tried comparing student submissions with answers from widely available AI tools, but the results rarely settled anything. I couldn’t prove that particular students had outsourced their work; by the time I left teaching earlier this year, though, I suspected at least 90% of my online students were using AI for some or all of their assignments.
In-person classes offered clearer moments, but not easy answers. One student gave a presentation, then couldn’t explain the thinking behind it when classmates asked basic questions. In a private conversation, the student admitted to using ChatGPT for part of the assignment and promised to stop. The same thing happened on the next assignment, followed by another promise.
I asked my supervising dean whether the case belonged before the academic integrity board. The decision was left to me; the university and college were still working out their position as AI spread through classrooms. ASU’s stated approach to AI includes a commitment to “using AI not only to address educational challenges, but to do so in a way that respects and uplifts all learners.” That’s an aspiration, but it doesn’t answer the immediate question facing a teacher who can’t tell whose work is on the page.
Learning can’t be reduced to a grade
People cheat at plenty of things: taxes, board games, even the occasional contest that feels stacked against them. School can be competitive, too. But learning isn’t a game you win by reaching the right answer. Its value is the ability to use what you’ve learned when you need it, as naturally as speaking a language or opening a door.
AI can look like a shortcut to good grades and plausible answers, especially when a busy teacher has little chance of identifying every generated passage. But if students hand over the work of learning, they may get a polished result without gaining the fluency that result is supposed to demonstrate. They also risk damaging their credibility when they can’t explain what they’ve submitted.
Here’s the uncomfortable part: I often couldn’t establish what had happened, and the tools available to me weren’t enough to settle the question. Yet the pattern was hard to ignore. Until people can connect their brains directly to computers, AI remains a crutch—and educators can’t keep treating the grading of ordinary homework as if it reliably measures student learning.
Rachel Kane is a contributor and former senior editor. She is a Los Angeles-based freelancer who has written and produced content for nearly two decades across technology, fashion, health, lifestyle, entertainment and education. The article was edited by Corinne Reichert and published October 26, 2026. The lead image shows a young woman using an AI assistant for homework; credit: Francesco Carta fotografo/Getty Images.
About the author
Rachel Kane is a Los Angeles freelancer who has spent nearly 20 years producing work on technology, fashion, health, lifestyle, entertainment and education. She teaches at Arizona State University’s Walter Cronkite School of Journalism and Mass Communication as a Professor of Practice. She also jokes that politeness toward chatbots might help avert the singularity, and admits to spending too much time tuning Midjourney prompts. Read her full bio or email CNET’s editors.
Her recent CNET stories include “AI Killed the Homework Era,” a test of AI agents for managing relationships and an experiment asking chatbots to plan her days and nights. The through line is pointed: AI tools can take on tasks, but that doesn’t mean they understand what people value.
More from CNET News
Elsewhere on CNET: AT&T’s $177 million data breach settlement payments, Oura’s renewed patent suit against Ultrahuman, rising Halloween candy prices and the lawsuit accusing Anthropic and Suno of unlawfully using songs by Madonna, Johnny Cash and Joni Mitchell. See all CNET news.