Survey reveals teachers fear AI’s impact on students’ critical thinking
Artificial Intelligence has already become routine in many industries, but when the conversation reaches classrooms, the tone shifts. A recent survey by NPR in partnership with Ipsos polled 545 K-12 teachers across the United States and delivered results that deserve attention: nearly 3 in 4 educators believe AI has bigger implications for education than previous innovations like the internet or computers. This is not hype, and it is not panic. It is a real concern from the people on the front lines every day, trying to balance the enormous potential of this technology with the risks it poses for student development.
The picture the survey paints is complex and full of nuance:
- Many teachers already use AI to make their own work easier
- But most are worried about what that same technology is doing to students’ ability to think for themselves
- And schools, for the most part, still don’t really know what to do about any of it
According to Mallory Newall, senior vice president at Ipsos, teachers feel AI will fundamentally reshape the future of education. She pointed out that educators have serious concerns about the technology’s impact on the relationship between teachers and students, and also on how students relate to each other.
If you think the debate about AI in education is still a thing of the future, this survey shows it has already arrived, and teachers are in the middle of the storm without much of a compass. 🧭
What the numbers actually say
When a nationally representative survey polls hundreds of teachers and nearly all of them point in the same direction, it is worth paying attention. The NPR and Ipsos survey did not ask whether AI was good or bad. It asked about impact, and the answer was practically unanimous among the educators consulted. The perception that Artificial Intelligence represents a deeper turning point than the arrival of computers in schools during the 80s and 90s, or even the internet in the 2000s, says a lot about how this technology is being felt in day-to-day classroom life. We are not talking about a tool that makes one task here or there easier. We are talking about something that changes the student’s relationship with the very process of learning.
Another standout finding: 6 in 10 teachers surveyed say they have already used AI to help with work-related tasks. They use the technology to create lesson plans, generate assessment questions, adapt content, and save time on administrative duties. However, the majority of those educators, about 63%, say the time savings from AI amount to two hours or less per week. In other words, the benefit is real, but it is not as dramatic as many people imagine.
On the other hand, student use is still relatively limited inside classrooms. Just over half of the teachers say their students simply do not use AI in class. About 2 in 5, however, report that students already use the technology at least once a week. This shows that adoption is still uneven, but the growth trend seems inevitable.
This is not a case of educators rejecting technology. Quite the opposite. What is at stake is a very important distinction: using AI as professional support for someone who already has a well-formed skill set is completely different from putting that technology in the hands of students who are still building the foundations of reasoning, argumentation, and the ability to solve problems independently. That difference is at the core of everything the survey reveals.
More of a teacher tool than a classroom resource
One interesting takeaway the survey makes clear is that AI has been functioning far more as an assistant for teachers than as a direct pedagogical tool in the hands of students.
Michele Naber, a biology teacher with many years of experience at El Toro High School in California, shared that she allows AI use in certain specific lessons. Her goal is not to let the tool do the work for the students, but rather to teach them how to ask better questions of chatbots and, most importantly, how to verify whether the generated answers are actually correct. In a typical exercise, she asks students to prompt ChatGPT for the physical characteristics and habitat of a particular animal and then compare the information against reliable sources. The result? Students realize that AI still gets things wrong, and quite often.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Naber also shared that using AI to generate multiple-choice test questions was a meaningful change in her routine. A task that would normally take over an hour to do by hand can now be completed in about five minutes. That kind of productivity gain is real and helps educators spend more time on activities that truly require a human presence.
Meanwhile, Joann Purcell, a math teacher and instructional coach at a school in the Chicago suburbs, found AI to be a helpful ally for creating professional development activities for her colleagues. However, she does not use the technology with her students and reports that, when it comes to generating math problems, AI is still not reliable enough. According to Purcell, reviewing the generated questions for errors ends up taking so much time that it is easier to just write the problems from scratch. 😅
Critical thinking at risk: educators’ biggest concern
If there is one topic that unites virtually every teacher in the survey, it is the concern about critical thinking. The numbers are clear: 54% of teachers believe AI makes it harder for students to develop critical thinking skills. The logic is easy to follow. When a student uses an AI tool to generate a ready-made answer, they skip the most important part of learning, which is the process of thinking, making mistakes, adjusting, arguing, and arriving at a conclusion on their own.
It is precisely in that process, often laborious and even frustrating, that the brain develops skills that go far beyond the content itself, like the ability to analyze information, identify contradictions, ask better questions, and make well-informed decisions.
Christa Corricelli, a special education teacher at a school near Boston, put it bluntly. She believes students who are already intrinsically motivated to think critically, that very small percentage of any class, probably will not be as affected. But for the vast majority, who do not have that disposition naturally, she fears that critical thinking skills will atrophy over time.
On top of that, 55% of the teachers surveyed believe AI primarily serves as a shortcut for students to avoid doing more work. The problem is that this shortcut looks very appealing, especially to young people who grew up in a digital environment where everything is fast and instant. Asking a tool to write an essay, solve a logic problem, or summarize a book chapter is tempting, and the output usually looks good enough to fool even traditional assessments. But what stays with the student after that? 🏃
Naber, in California, feels a deep responsibility to teach her students that humans should always question and verify what AI generates. In her view, if people stop questioning what the technology tells them, they can be led to believe anything, and that possibility genuinely scares her.
The exception that proves the potential
It is not all concern, though. Ellie Rodriguez, a special education teacher at a school near Palm Beach, Florida, offered an important counterpoint. She shared that AI can be especially useful for students with disabilities. One of her students, who is on the autism spectrum, recently used an AI tool to get help with an assignment he would not have been able to complete on his own. Rodriguez praised the student for his initiative, comparing the use of AI to using an encyclopedia or a library book as a resource to find answers.
Still, Rodriguez acknowledges that the same technology that helps this type of student can hurt those who are capable of doing the work on their own. She and her colleagues, including English language arts teachers, are deeply worried about AI’s impact on students’ ability to think independently.
Trust between students and teachers is being eroded
One of the most alarming findings in the survey is this: nearly 6 in 10 educators say AI is eroding the level of trust between students and teachers. Mallory Newall at Ipsos called the erosion of that trust one of the biggest red flags found in the data.
And the problem gets worse when combined with another finding from the survey: 70% of teachers believe the public perception of them has worsened in recent years. In other words, educators are trying to navigate incredibly complex challenges in an environment that is already marked by distrust.
In practice, the erosion of trust is driving concrete changes in how teachers do their work:
- About 4 in 10 have started requiring that more assignments be completed by hand
- Another 4 in 10 have increased the amount of work done inside the classroom, under supervision
Naber, in California, gave an example that illustrates this new reality well. For years, she offered extra credit to students who participated in beach cleanups and habitat restorations outside of school hours. The proof was simple: just a photo. But after her own son showed her how easy it is to use AI to create a fake image of a registration table at one of those events, she had to scrap the practice entirely. There was no longer any way to verify the authenticity of the photos. She also revamped her curriculum so that all lab work is done in class, in her presence, and homework now counts for much less of the final grade.
Josh Kauffman, who teaches English to seventh graders at a public virtual school in Alabama, faces an even bigger challenge. Because the school operates entirely online, he cannot simply require more in-person work. Instead, he tries to convince his students that there is value in their own writing, typos and all. For Kauffman, it is better to deal with students’ authentic mistakes than to spend time wondering how much they are leaning on technology to do the work.
Purcell, in Illinois, has a different take. She does not necessarily think AI has eroded trust, since students have always found ways to get around assignments long before AI existed. In her view, teachers need to get creative with the technology and have students think alongside it, the same way they would with any other tool. 🤔
Schools are leaving teachers without guidance
Perhaps one of the most worrying aspects the survey revealed is the lack of institutional support for educators. The data shows that many teachers are having to adapt to the AI era essentially on their own, without clear guidelines from their schools or districts.
Among teachers whose schools provide AI software, only 35% say there is a formal policy on technology use by educators. That means institutions are handing out the tools without establishing clear rules for how to use them, a scenario that breeds confusion and inconsistency.
About half of all teachers surveyed say their schools have not offered any guidance on AI, or they are not even sure what the current guidance is. Only about 4 in 10 say their schools offer professional development or training related to the technology.
Rodriguez, in Florida, shared that she has not received any training on AI and would really like the opportunity. In her view, schools need to teach educators how to apply the technology in positive ways to what they do and, more importantly, to how they teach.
Kauffman agrees and warns that not enough attention is being given to how to teach what is already being taught differently, taking into account the flexibility and resources AI can offer.
Corricelli, near Boston, says she is not entirely surprised by the lack of training. She notes that schools tend to be slow to adapt to change, and that has been an enormous challenge for educators. In her words, everyone is basically just trying not to drown. 😬
Responsible use needs to be taught in schools
One of the clearest signals the survey sent is this: nearly 8 in 10 teachers believe schools should teach the responsible use of AI. Newall at Ipsos called this an unmistakable message that educators recognize the enormous implications of the technology for education and know it is not going away. The time to act, she said, is now.
The answer is not to ban Artificial Intelligence from education. That would be both impossible and counterproductive, since we are talking about a technology that will be present in virtually every career and context these students will face in the future. What the best-prepared teachers are trying to do is create a mediated, conscious, and pedagogically guided relationship with these tools. That means teaching students to use AI as a starting point for investigation, not as a final destination. It means using the results generated by the tool as objects of critical analysis, questioning what was said, identifying gaps, and comparing with other sources.
The concept of responsible use of AI in schools needs to be built collectively and institutionally. It cannot depend solely on the individual effort of each teacher. That involves creating clear guidelines on when and how students can use these tools, offering ongoing professional development for educators, and adapting assessment methods so they genuinely measure a student’s reasoning rather than just the final product submitted. When young people understand the purpose behind intellectual effort, they tend to make better choices, even when they have access to technological shortcuts.
Families and society at large also have an important role to play in this equation. The pressure for fast results, high grades, and measurable performance often pushes students toward the very shortcuts that undermine the development of critical thinking. Rethinking what we value as quality learning is a conversation that needs to happen outside the classroom too. Teachers cannot drive this transformation alone, and the NPR and Ipsos survey makes that quite clear. 💡
What to expect going forward
The trajectory of Artificial Intelligence in education is still being written, and the next few years will be decisive in defining the paths schools choose to take. Some countries and states are already moving forward with regulations and specific guidelines for AI use in educational settings, and those initiatives serve as a reference for anyone still trying to figure out how to act. In Brazil, the debate is also growing, though more slowly and unevenly depending on the region and type of institution. What is certain is that ignoring the topic is no longer a viable option, because the technology is already inside students’ backpacks, in the phones they carry to school every day.
The teachers who participated in the American survey are not asking to turn back the clock. They are asking for support, guidance, and a serious conversation about how to integrate AI without giving up what matters most in human development. That is a legitimate demand that deserves real attention from administrators, researchers, and policymakers around the world. Technology moves too fast for education to sit still waiting for perfect consensus, but that does not mean anything goes or that there are no choices to be made. On the contrary, the choices made right now will shape entire generations.
At the end of the day, what the survey puts on the table is a fundamental question: what kind of skills do we want students to develop, and how can AI serve that goal instead of replacing it? That question does not have a simple answer, but it needs to be at the center of every decision involving technology and education. And who better to contribute to that answer than the teachers who live this reality every single day? 🎓
