While schools have been grappling with the student use of artificial intelligence in classrooms, a growing shift is happening on the other side of the desk. Teachers are increasingly turning to AI tools for grading, lesson planning, and other responsibilities. As platforms like ChatGPT and Google Gemini are shifting the way educators work, new concerns arise over fairness and regulation.
In March, New York City Public Schools released its first official guidance on AI use in classrooms, which listed out what AI is and what it is not. The document distinguished AI from a “thinking, reasoning, or conscious being,” and clarified that it cannot “exercise judgment the way people do.”
The guidance categorizes AI risks and benefits by using the analogy of a traffic light to clarify what uses of AI are and are not allowed. Under the red category, it is strictly prohibited for educators to use AI to make decisions about student grading, placement, discipline, and counseling. Under the yellow category, student use of AI for educational purposes is permissible, but only with active educator oversight. The yellow category also lists uses for educators but these are ones to be used with caution. For instance, AI can be used to generate translations for critical communications and translations to support multilingual learners, both of which must be reviewed by professionals prior to distribution. The green category includes approved uses for educators such as brainstorming lesson ideas, organizing or summarizing information, and translating non-critical school communications for families and communities who prefer a language other than English.
The guidance also explains that AI can potentially be a supportive tool in classrooms but should not replace original thoughts or writing.
Social Studies teacher Francis McCaughey said that as AI becomes more common in classrooms, clearer expectations for its use are needed. He said, “I think [AI] can be a great tool for teachers but much like we expect of students, some guidelines are needed.” He said that verifying the accuracy of AI-based materials is especially important in his social studies courses, where students often require authentic historical sources in preparation for exams.
The reliance on these sources makes it important for teachers to verify anything produced with AI assistance. Though AI is often meant to be a time-saving tool, he said, “…the AI bot sometimes invents documents that are an amalgamation of other online sources but not actual documents… Now instead of time spent searching, I spend time ensuring that AI is providing authentic documents.”
While the NYCPS guidance provides clear expectations for how teachers should use AI and the importance of reviewing AI-generated work, some students say that their classroom experiences do not reflect consistent practice of those expectations.
The NYC Schools Guidance states that “All AI tools must be used with human oversight and review,” however many students feel that it is unclear when that standard is being met.
A junior said she has experienced receiving an AI-generated assignment. When asked how she knew it was AI-based, she said there was an automated pre-written note introducing AI-generated content at the top of the assignment indicating its use. “I felt like it was unfair because we get punished heavily for the same thing they’re openly doing,” she said.
One freshman said they learned about the limitations of AI feedback during Writing Process, where their teacher showed them the difference between AI feedback and human feedback. “[AI feedback] was so much more different than a teacher’s feedback and completely inaccurate in terms of the point system and rubric requirements,” she said. “A teacher would have more accurate knowledge in the rubrics and requirements.”
National data suggests that teacher use of AI is rising rapidly. According to the EdWeek Research Center data, the percentage of teachers using AI-driven tools in their work nearly doubled between 2023 and 2025. In 2023, 34 percent of teachers said they used AI at least “a little,” but by 2025, 61 percent said they did.
In New York public schools specifically, where teachers manage large class sizes and assignments, AI is emerging as a source of support. An August 2025 report from the New York City Comptroller’s Office cited surveys showing that nearly half of teachers and students use ChatGPT at least once a week, up from 27 percent the year prior.
The NYCPS guidance came out just as AI tools designed specifically for educators became more easily available. Google Gemini, which is now integrated into Google Workspace across apps like Gmail, Docs, Sheets, and Drive, can generate lesson outlines and classroom materials. Companies such as Teq, which works alongside schools to provide instructional technology, have also promoted training for applying AI tools to tailor assignments. These tools are embedded into teaching platforms and provide easier access to teachers, increasing the reliance of technological support in classrooms.
Though these AI tools can improve efficiency, using them raises ethical concerns. Some worry that the technology may introduce bias and weaken transparency, particularly when used for tasks that traditionally rely on human judgment, such as grading student work or writing recommendation letters.
Some critics of AI use in educational settings argue that schools should move cautiously. Kelly Clancy, founder of Parents for AI Caution in Educational Spaces, said parents should consider AI’s potential to hinder the learning process.
She said that if teachers use AI tools, it can undermine the goal of teaching, which “isn’t to produce a perfect product— it’s all of the learning that happens during the process.” As a result, Clancy said that schools should clearly disclose when AI tools are used in classroom materials or feedback, as students “have a right to know whether they are being taught and evaluated by a human or a machine.”
It is difficult to monitor if teachers are using AI as outlined in the traffic-light document, and students are uncertain about how much AI teachers are using in their classes. Junior Ananya Detering said teachers should always inform students when AI is involved, because “it affects the validity of what is provided.” She said, “AI feedback feels different from teacher feedback, because AI is unable to annotate specific parts of a person’s writing, and in my experience, AI has always given more vague responses.”
Ananya said that heavy reliance on AI can make the learning process feel less personal, and “it isn’t the teacher teaching anymore.”
Similarly, sophomore Hifsa Khan said, “I’ve noticed AI use in [one class], especially with our recent slideshow project. The feedback we got didn’t really feel like a person actually read through it. It felt as if our teachers just put each of our slides into a generator and asked it to point out the pros and cons, then copy and pasted them as a private comment on our work.”
Hifsa said that she has noticed a pattern with AI-generated feedback. “It feels super repetitive. AI loves ‘fluff’ so it’ll tell you your structure is ‘effective’ or ‘strong’ without actually pointing to a specific line you wrote,” she said. The vague feedback made it difficult to implement necessary changes in her work.
The use of AI in writing student recommendations is also a potential problem, because it is a task that requires personal knowledge of students to convey authenticity. An EdWeek report found that about one-third of teachers, surveyed by the education research group foundry10, have used AI tools to help draft recommendation letters. Given that teachers typically write dozens of time-consuming letters a year, the convenience of AI tools is tempting. For students, however, this practice raises concern, as teacher recommendations play a significant role in their futures.
In a statement emailed to The Classic, New York City Public Schools did not directly address whether AI use in college recommendation letters are permitted. The statement said that educators should remain “fully accountable for all content they produce” and that “AI should enhance, not replace, professional judgment and expertise.”
One junior said, “I think recommendation letters should be personal… and the use of AI abolishes this trust between me and the mentor I trusted to write this recommendation.” She added that knowing AI was used “would lead [her] to doubt the quality of [her] recommendation.”
As teachers continue experimenting with AI, questions about transparency remain unsolved. Even with the guidance outlining permissible AI use, the rapid growth of AI generated feedback and reliance raises concern on how human judgment and machine assistance is maintained. Without any way to monitor or disclose how AI is used, the distinction depends on teachers.
Correction: An earlier version of this article referred to a Writing Process class providing students with AI feedback on essays. This article has been updated to include additional context to show that the purpose of the lesson was to show the limitations of AI feedback.
