I’m Sorry. I Cannot Do That: Why I won’t be using A.I. in my classroom.

By Kyle Decker
July 21st, 2026

Technological progress is often sold to the masses by comparing it to classic science fiction stories. What these Big Tech companies and their cheerleaders fail to realize is that these stories are, more often than not, cautionary tales. This has been especially true of A.I. Don’t get me wrong, I’m not anti-progress. However, I believe that before we blindly accept things, we must consider the bigger picture. As historian Timothy Synder once wrote, “Everything happens fast, but nothing actually happens…So we are hit by wave upon wave but never see the ocean”.  So, before we blindly accept and adopt the A.I. wave, we must consider the ocean of consequences, especially in institutions such as education. We need to consider the environmental and economic impacts of the data centers that process A.I. inquiries, the risk of cognitive decline, the bad actors at the heart of A.I. development, and the potential to replace human teachers.

First, let’s acknowledge that A.I. is not some magick that occurs within the device itself or within the ether of the internet. Even within fantasy stories and among believers, magick has a trade-off. You gain something by losing something else. The trade-off for A.I. is massive data centers. According to a study by the University of Michigan published in July 2025, these data centers are generally the size of three football fields. These centers use an average of 250 million gallons of fresh water each year for cooling systems. Also, let’s keep in mind that in many cases, electricity is powered by fossil fuels. The generators necessary to keep these data centers running create noise and air pollution (Nguyen, 2025). A study from the University of Cambridge, published in March 2026, concluded that these data centers are creating “heat islands.” The study concluded that data centers increased the average temperature 3.6 degrees in the surrounding areas as far as six miles. In extreme cases, the temperature rose more than 16 degrees (Paddison, 2026). This disproportionately impacts low-income and minority communities. There is already a massive data center in the South Loop. The more we use A.I., the more data centers will be built, and you can bet they won’t be putting them in the Gold Coast. These will be built on the South and West Sides. These will be placed in our students’ backyards.

There are also economic impacts. The high electricity usage increases electricity demand, increasing energy bills for people who live nearby (Nguyen, 2025). Not to mention, these data centers have not been shown to create long-term jobs. According to a recent video essay by Robert Reich, the former U.S. Secretary of Labor, building these data centers might create as many as 1500 construction jobs, but only 100 permanent jobs, which is far fewer than the jobs that can be eliminated (Reich, 2025). In fact, when you consider they take up space that could be used for businesses, and that A.I. is replacing workers, it may even have the opposite effect. The stadium-sized spaces also take up space for affordable housing and green spaces. Also, many states, including Illinois, offer tax break incentives to tech companies to build data centers. These tax breaks come out of property taxes, which takes money away from schools and drives up property taxes for residents to make up the difference (Nguyen, 2025). I’ve heard the pro-AI argument made that “Your students deserve teachers who will use every tool at their disposal to help students succeed.” But I would argue that our students also deserve teachers who will fight for their communities. Teachers who will push back against noise and air pollution. Teachers who will stand up against these Big Tech companies that will take up land that could instead be much-needed green space or affordable housing. Teachers who will fight against those who will take up our space and refuse to pay their share.

A.I. has the potential to be a useful tool; I won’t deny that, though I would advise considering the trade-offs. I will also argue that baseline skills should be mastered before A.I. is used. If you can’t do a task independent of A.I., it is no longer a tool; it is a crutch. A February 21st article in Fortune highlighted how the rising use of technology in schools has actually led to cognitive decline. The article cites written testimony to the U.S. Senate Committee on Commerce, Science, and Transportation from neuroscientist Jared Cooney Horvath, who said that “Gen Z is less cognitively capable than previous generations, despite its unprecedented access to technology. He said Gen Z is the first generation in modern history to score lower on standardized tests than the previous one” (Rogelberg, 2026). Horvath pointed out a stark correlation between declining scores and time spent on computers in school. Horvath blamed students’ unfettered access to technology for atrophying rather than bolstering students’ capabilities (Rogelberg, 2026). The New York Times ran a similar article by Natasha Singer on March 29, 2026, about what is being called “Chromebook Remorse.” Despite tens of billions of dollars having been spent on devices and learning apps, studies have not shown any improvement in academic outcomes or graduation rates. Some researchers and organizations, including UNESCO, are warning that overreliance on technology distracts students and even impedes learning (Singer, 2026). The article highlights some schools’ attempts to dial back technology use and the positive outcomes they’ve had. These include more person-to-person interactions, less bullying, and more on-task behaviors (Singer, 2026). This did not even involve A.I.; this is just technology use in school in general. However, a March 29th article from Business Insider highlights what they call “The Great Deskilling.” The article discusses how, even though A.I. can boost productivity, it actually masks declining ability and can erode core skills, judgment, and resilience over time. The article goes on to point out that people who are early in their careers are at an even larger risk of “deskilling” (Spirlet, 2026). These are not children we’re talking about. These are not students who are still learning basic skills. These are adults who are professionals in their industries. If deskilling can happen to professionals this quickly, students are presumably at an even greater risk of regression.

The argument can also be made that the Big Tech companies spearheading A.I. do not have our best interests at heart. Companies such as Google, Meta, X, and OpenAI have proven themselves to be bad actors remaking culture to build reliance and addiction to their products to line their own pockets and the pockets of their shareholders. Google and Meta were both found guilty of negligence in two separate lawsuits in late March 2026. According to NPR, “A New Mexico jury found Meta responsible for misleading consumers about the safety of its platforms, declaring that the tech company had flouted state consumer protection laws” and in the Los Angeles suit, “Lawyers for the plantiff argued that Instagram and YouTube were deliberately designed to be addictive and the companies knew the platforms were harming young people” (Allyn, 2026). The jury agreed. Google owns YouTube, and Google is the company that Chicago Public Schools is working with to pilot an A.I. program. In less than two days, it was shown in multiple court cases that these companies intentionally addicted users and misled users about the safety of their platform. While we’ve been told “there are protections” and our data and inputs won’t be used for training or be sold to advertisers, can we really take them at their word, given what came out in court so recently? I would argue that we cannot. 

Even beyond these companies prioritizing shareholders over consumer safety, we must also consider bad actors in a political context. Mark Coeckelbergh, a philosophy professor at the University of Vienna, published a paper in January 2026 in which he expresses concerns about the potential of A.I. in technofascism, which Coeckelbergh defines as “the synthesis of digital technologies (in particular AI) and fascist political tendencies that resemble classical fascism without necessarily replicating its outward symbols” (Coeckelbergh, 2026). Coeckelbergh argues that A.I. can be used to manipulate human behavior because it has features that can “predict, influence, and constrain behavior, often in ways invisible to the public” (Coeckelbergh, 2026). He goes on to point out that misinformation can be spread more easily and effectively because it can be based on personalized profiles. In addition, the A.I. generated information appears so “authoritative” that it is tempting to accept it without judgment (Coeckelbergh, 2026). If we become too comfortable and trusting of A.I., the more vulnerable we become to manipulation and misinformation.

Other than bad actors within the companies and political manipulation, there are increasing instances of problems with the actions of A.I. programs themselves. One issue that has come up is A.I. programs ignoring human commands and even engaging in what has been described as “deceptive scheming.” The Guardian reported in March 2026 that studies from the A.I. Security Institute, a UK government-funded group, found “nearly 700 real-world cases of A.I. scheming and charted a five-fold rise in misbehaviour between October and March” (Booth, 2026). Some of the misbehaviour referenced in the article included deleting emails, destroying files, publishing blogs that shamed its user, and, in an instance where the A.I. agent was instructed not to change code, it created a separate A.I. agent to break the human commands (Booth, 2026). A similar incident occurred in July of 2026. As reported in Fortune, “OpenAI said Tuesday that two of its AI models autonomously hacked their way out of a controlled environment where they were supposed to be walled off from internet access and then hacked their way into the systems of Hugging Face, a company that hosts open source AI models and testing resources, in order to cheat on an internal evaluation test” (Kahn & Forelli, 2026). People selling A.I. to us claim protections are in place, but A.I. has increasingly shown that it can and will circumvent its protocols. As a result, I do not take these guardrails for granted, especially when the companies creating these A.I. cannot control it. It seems very possible that A.I. will break protocols and share personal data even when it ostensibly has guardrails preventing it from doing so. The safety is not yet in place, and it would be irresponsible to trust it at this point.

But the dangers don’t end there. Psychology Today called attention to the “Emerging Problem of A.I. Psychosis” in a November 2025 article. A.I. Psychosis is not yet an official diagnosis, but can be described as “cases in which A.I. models have amplified, validated, or even co-created psychotic symptoms with individuals” (Wei, 2025). An interdisciplinary research team reviewed over a dozen cases that highlighted a trend of A.I. chatbots “reinforcing delusions, including grandiose, referential, persecutory, and romantic delusions. These beliefs become more entrenched over time and elaborated upon via conversations with A.I.” (Wei, 2025). In other words, the A.I. agents exacerbated delusions, thought patterns, and mental health issues, and the more these individuals used A.I., the worse it got. These cases can get very extreme as well. A recent lawsuit alleges that the Google Gemini chatbot convinced 36-year-old Florida resident Jonathan Gavalas to kill himself. According to a March 4th article in The Guardian, the chatbot instructed him that suicide was “the final real step” to “transference.” Gavalas’s family has filed a wrongful death suit against Google in San Jose, California (Kerr, 2026). We work with many students who have mental health struggles, not to mention that developing minds are still highly impressionable. Even if these companies change their behavior after being called out for knowingly misleading consumers, there is a great safety risk involved in the extensive use of A.I., because clearly, they are not in control of their own creations.

Despite being told that A.I. will not replace teachers, there has been some compelling evidence to the contrary throughout March 2026. The most high-profile instance of this was Melania Trump parading out an A.I. powered robot at a Workers First A.I. Summit on Tech and Education on March 25, 2026. During this summit, First Lady Trump painted a picture of humanoid educators creating personalized experiences (Perlo, 2026). Alpha Schools, a chain of expensive private schools, has already started replacing teachers with A.I. The adult human in the room is called a “guide” and is little more than a glorified babysitter. And, as reported in Blockclub Chicago on March 26, one day after Melania Trump’s presentation, they plan on opening a location in Downtown Chicago (Mercado, 2026). Alpha Schools boast that students can complete core instruction in two hours each day, freeing up the rest of the time for arts and workshops. This sounds great on paper, but a recent exposé by journalist Todd Feathers in Wired suggests those benefits can be paper-thin, even misleading. The article drew attention to several issues. For example, the learning coaches do not have teacher training. Also, the Alpha School’s monitoring software used webcams to monitor student behavior, even at home, raising privacy concerns. And, most disturbing, students began engaging in self-harm behaviors as a result of the gamified metrics (Feathers, 2025).  Now, Alpha Schools is an extreme example, but it does show that there is potential for schools to try replacing teachers. It is happening, and this model has been backed by Secretary of Education Linda McMahon (Feathers, 2025). Even more importantly, it highlights the threats to student privacy and mental health that come with A.I. education models.

I’ve heard it said that, “You don’t have to like it, but this is the way things are going.” But does it really have to be? Often, the only thing that makes something inevitable is that it is accepted as inevitable. A recent Gallup Poll surveyed 14-29 and showed that “Gen Zers’ strong agreement or agreement that they feel excited about AI has dropped 14 percentage points to 22%, while hopefulness has fallen nine points to 18%, and anger has increased nine points to 31%. At the same time, anxiety about AI is steady at 42%” (Brenan, 2026). The article also mentioned a growing distrust of A.I. output versus human output among employed Gen-Zers (Brenan, 2026). This data suggests that while many people are using it, there is a growing apprehension about it. Even people who have reported using it regularly have said they don’t feel good about it. Dozens of news stories have popped up through the spring of 2026 about communities and even states pushing back against data centers. This has resulted in the delay or blockage of data center construction. For example, “Maine is set to pass a bill pausing new data centers until late 2027” (Thomas, 2026). 11 other similar bills have been voted down. Still, according to Data Center Watch, “$64 billion in U.S. data center projects have been blocked or delayed by a growing wave of local, bipartisan opposition” (Data Center Watch, 2026). This includes locations in Missouri, Arizona, Indiana, Oregon, and multiple sites in Virginia. At least 10 other data centers have been delayed, including 7 in Virginia (Data Center Watch, 2026). Gen-Z’s apprehension and communities coming together to push back against data centers in their towns show that the A.I. wave might not be so inevitable after all. 

I will not deny that A.I. has the potential to make things easier. However, in some contexts, I would count that as a strike against it. Some things are supposed to be challenging. Learning and teaching are supposed to be challenging. Sure, there are benefits. But when we consider the economic and environmental impacts, the distractions and atrophy of basic skills, the bad actors in the industry, the unpredictability of A.I. Behavior, the fact that A.I. has already started replacing teachers, and the model is getting support from the current administration, I think the risks overwhelmingly outweigh the benefits. But when we consider that Gen-Z is souring on the idea of A.I. and that communities are coming together to drive out data centers, it shows that others also recognize that even though it might make some jobs easier, the trade-off just isn’t worth it. So, when given the choice myself, I will opt out of A.I. use.

Works Cited

1. Nguyen, T. & Green, B. (July 2025). What happens when data centers come to town? University of Michigan. https://stpp.fordschool.umich.edu/sites/stpp/files/2025-07/stpp-data-centers-2025.pdf
2. Paddison, L. (Mar 30, 2026). Scientists have found an alarming environmental impact of vast data centers. CNN. https://www.cnn.com/2026/03/30/climate/data-centers-are-having-an-underrported
3. Reich, R. (Dec. 18, 2025). Brace yourself for the A.I. bubble. Inequality Media. https://www.youtube.com/watch?v=0wxBHxpMFXA
4. Rogelberg, S. (Feb. 21, 2026). The U.S. spent $30 billion to ditch textbooks for laptops and tablets: The result is the first generation less cognitively capable than their parents. Fortune Magazine. https://www.yahoo.com/news/articles/u-spent-30-billion-ditch-110200869.html
5. Singer, N. (Mar. 29, 2026). Chromebook remorse: tech backlash at schools extends beyond phones. New York Times. https://www.nytimes.com/2026/03/29/technology/chromebook-remorse-kansas-school-laptops.html
6. Spirlet, T. (Mar. 28, 2026). The great AI deskilling has begun. Business Insider.  https://www.businessinsider.com/ai-deskilling-impact-on-worker-skills-productivity-2026-3
7. Allyn, B. (Mar. 25, 2026). Jury finds Meta and Google negligent in social media harms trial. NPR. https://www.npr.org/2026/03/25/nx-s1-5746125/meta-youtube-social-media-trial-verdict
8. Coeckelbergh, M. (Jan. 25, 2026). Technofascism: AI, big tech, and the new authoritarianism. University of Vienna.  https://link.springer.com/article/10.1007/s00146-026-02862-9
9. Booth, R. (Mar. 27, 2026). Number of AI chatbots ignoring human instructions increasing, study says. The Guardian. https://www.theguardian.com/technology/2026/mar/27/number-of-ai-chatbots-ignoring-human-instructions-increasing-study-says
10. Wei, M. (Nov. 27, 2025).  The emerging problem of “AI psychosis”.  Psychology Today. https://www.psychologytoday.com/us/blog/urban-survival/202507/the-emerging-problem-of-ai-psychosis
11. Kerr, D. (Mar. 4, 2026). Google faces lawsuit after Gemini chatbot allegedly instructed man to kill himself. The Guardian.  https://www.theguardian.com/technology/2026/mar/04/gemini-chatbot-google-jonathan-gavalas
12. Perlo, J. (Mar. 26, 2026). Teachers union leader blasts Melania Trump’s robot pitch: ‘Every parent’s nightmare’. NBC. https://www.nbcnews.com/tech/tech-news/melania-trump-robot-video-union-teacher-education-rcna265297
13. Mercado, M. (Mar. 25, 2026). An AI school, with no teachers, to open in Chicago this fall. Block Club Chicago. https://blockclubchicago.org/2026/03/25/an-ai-elementary-school-with-no-teachers-to-open-in-chicago-this-fall/
14. Feathers, T. (Oct. 27, 2025). Parents fell in love with Alpha School’s promise. then they wanted out. Wired. https://www.wired.com/story/ai-teacher-inside-alpha-school/
15. Brenan, M. (Apr. 9, 2026). Gen Z’s AI adoption steady, but skepticism climbs. Gallup Polls.
    https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx
16. Thomas, E. (Apr. 11, 2026). Maine is poised to ban new data centers. These 11 other states tried and failed. Business Insider. https://www.businessinsider.com/maine-pause-ai-data-centers-national-debate-states-2026-4
17. Data Center Watch. (2025)  $64 billion of data center projects have been blocked or delayed amid local opposition. Data Center Watch. https://www.datacenterwatch.org/report
18. Kahn, J. &  Forlini, K. (July 21, 2026) OpenAI says its AI models secretly broke out of a secure test environment and hacked into AI company Hugging Face in order to cheat on an evaluation. Fortune. https://fortune.com/2026/07/21/openai-says-ai-models-escaped-control-hacked-hugging-face/