A Year to Get This Right
This article has been published on Substack: https://erikatwani.substack.com/p/a-year-to-get-this-right
On September 2, I sat with about 40 browser tabs open, each one with a different take on the new NYC policy. Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels had announced New York's one-year moratorium on generative AI in classrooms through 8th grade (New York City Mayor's Office, September 2, 2026).
A coalition mostly made up of parents who had advocated for a two-year moratorium welcomed the new rules as a positive step but warned that the exemptions could still jeopardize student safety (Elsen-Rooney, September 2, 2026). The teachers' union approved the new screen time limits but pointed out that the AI regulations leave important gaps unaddressed (CBS New York, September 2, 2026). A former deputy mayor criticized the blanket restrictions on younger students, deeming them inappropriate for a city where graduates must be workforce-ready (Governing, September 2, 2026). Overall, opinions mix relief and caution, with most falling into that nuanced space.
All those concerns deserve serious consideration, starting with parents. When parents see their children ask a chatbot to complete a paragraph they could be working on themselves, it’s understandable to worry that fundamental skills are being lost. Mamdani and Samuels share this concern, and they are right.
Protecting a young person's ability to think and maintain the connection between students and teachers is crucial. I have dedicated 15 years to developing an argument around these priorities. Teachers and school leaders share similar concerns. Over the last two years, many have tried a variety of AI tools without much opportunity to fully understand them, only to be told to discard these tools just as swiftly. I believe nobody's concerns in this debate are unjustified.
Looking at where AI succeeds with young students shows that progress is deliberate. Certain districts are already seeing real improvements, which is no accident. Schools that excel in this area have a common characteristic: they always supervise AI interactions with students. Instead of simply providing a chatbot and hoping for the best, teachers or tutors actively guide the process, shaping how the tool is used in real time.
This pattern shows that AI helps young learners the most when a caring adult stays involved. It provides less benefit if students use the tool without guidance. This emphasizes the need for guided use, exactly what New York's pilot programs are designed to test, ideally with a teacher present and paying attention.
I've maintained from the beginning that skills are more fundamental than tools, and agency precedes AI. No opinion piece, including this one, can change a policy the city has already enacted for a year. Instead of criticizing the moratorium itself, why not support those implementing it to get it right? If a year passes without structural changes, we both know the likely outcome: schools revert to pre-AI instruction methods, parents remain just as dissatisfied because their concerns aren't addressed, and students keep relying on chatbots for homework and using AI without the necessary critical thinking.
Here is what learning looks like when agency precedes tools: Imagine a typical guided activity that doesn't require a distant classroom. Visualize students working with the city's 311 data on rat sightings. They independently map the blocks with the most complaints and discuss trash pickup schedules with building supers. After laying this groundwork, they use AI to compare sightings with the city's containerization schedule, identify lagging areas, and develop a draft proposal for their district's timeline to present to their City Council member. The AI didn't file complaints or count rats. It simply extended the students' initial efforts, much like a microscope that enhances sight for a specific purpose. This is the ideal form of augmented intelligence, and any city council office could approve such a project tomorrow.
This is why I believe agency should be prioritized, even before students learn AI literacy or how to craft effective prompts. Agency involves self-awareness focused on your own thinking, such as recognizing when you're guessing versus when you genuinely know something. It also includes the willingness to persist with difficult problems rather than passing them on to others at the first block.
A self-trusting mind is more likely to take risks, and embracing the possibility of being wrong can be seen as a form of creativity. This creativity often leads to better questions than those provided outright to students. In my new book, The Einstein Mindset, I refer to inquiry as the new literacy, emphasizing understanding which questions are worth pursuing rather than solely seeking correct answers. This development is a one-way process: teachers can create an environment that fosters curiosity, but the student must ask the questions.
Two additional concerns require clear answers before concluding. First, data privacy is a significant issue. Any AI tool a district adopts should meet the same standards as the 38 programs flagged by the city this month (Duffy, September 2, 2026) and be tested against student privacy standards before classroom use, not afterward.
Second, equity is crucial. Wealthier districts can afford to hire consultants to develop carefully supervised AI programs, while resource-strapped schools may simply distribute an app and hope for the best. This disparity is real and needs to be acknowledged. That’s why implementing this well at the scale of New York is so important. If the sequence prioritizes agency first, with AI second, and is supported by public funds and oversight, it creates a model that smaller districts can emulate, rather than a patchwork where only the wealthiest schools succeed.
Chancellor Samuels, if this message reaches you, consider this year an opportunity and a runway for progress. Integrate agency pedagogy into learning practices now, while the challenge of reintroducing AI is still a year away. When AI is eventually reintroduced, make it conditional. Demand demonstrated real learning autonomy before granting access, like how you wouldn't give a lab tool to a novice who isn't yet capable of handling it safely. By doing so, this moratorium becomes a strategic foundation for reintroducing AI on New York's terms, on a timeline no other district has dared to pursue.
To everyone reading this — parents who sent me different articles each day this week, and teachers preparing for how things will unfold — I ask the same of you. Don't spend this year waiting for the city to decide. Instead, select one classroom or a small project, like the one above, and start cultivating a sense of agency now, whether at home or in whatever part of the school day you can influence. A year from now, when the moratorium ends, I want New York's students to be prepared to use the tools they will regain, rather than starting from scratch. This is the story I hope to tell next, and I’d love your help in writing it.
References
1. New York City Mayor's Office. (2026, September 2). Mayor Mamdani and Chancellor Samuels Put Students First with Nation's Broadest Generative AI Moratorium in Schools.
2. Duffy, C. (2026, September 2). Nation's largest school district bans AI in the classroom through 8th grade. CNN Business.
3. Elsen-Rooney, M. (2026, September 2). NYC's new school AI ban explained: What students and teachers need to know. Chalkbeat.
4. CBS New York. (2026, September 2). NYC officials ban AI in the classroom until high school.
5. Governing. (2026, September). New York City Put Limits on AI in Schools. Are They Enough?
