As another academic semester begins, something has changed in the conversation about artificial intelligence. For the last several years, the dominant discussion surrounding generative AI has centered on capability: what these systems can do, how quickly they will improve, which occupations they will transform, and which companies will ultimately lead the market. Increasingly, however, the conversation is shifting from what AI can do to what AI may be doing to us. That distinction matters because the questions surrounding artificial intelligence are becoming as much social and educational as they are technological.
Recent reporting from The Wall Street Journal, Financial Times, and Futurism paints a remarkably consistent picture despite approaching artificial intelligence from very different perspectives. Public opposition to AI infrastructure is growing, trust in the companies developing these technologies remains weak, and educators are increasingly concerned that students may be outsourcing intellectual work rather than augmenting it. At the same time, businesses continue adopting AI and increasingly expect employees to understand how to work effectively with it. Welcome to what I would call the Semester of Our Discontent.
It would be easy for advocates of artificial intelligence to dismiss this growing backlash as technophobia, but I believe that would be a mistake. It would be equally misguided for higher education to respond by attempting to place artificial intelligence back into a box that has already been opened. The more difficult task is determining how society can gain the benefits of artificial intelligence without surrendering important elements of human judgment, autonomy, and intellectual development. In education, that means learning to distinguish between AI that expands human capability and AI that replaces human engagement.
Simonetti and Bobrowsky (2026), writing in The Wall Street Journal, describe major technology companies scrambling to address growing resistance to the massive data centers required to support artificial intelligence. Companies including OpenAI, Meta, Microsoft, and Amazon are investing heavily in community outreach as residents raise questions about electricity consumption, water usage, environmental impact, employment, and the speed with which data center developments are being approved. These are no longer abstract concerns about a technology that exists somewhere in the cloud. Artificial intelligence now has a physical footprint, and communities are beginning to ask what they receive in return for supporting it.
Gallup found that seven in ten Americans oppose the construction of an AI data center in their local area, with 48% strongly opposed (Jones, 2026). Those concerns include energy consumption, water use, pollution, increased utility costs, and changes to local quality of life. Public distrust also extends well beyond infrastructure, as a Bentley University and Gallup survey found that only 31% of Americans expressed at least some trust in businesses to use AI responsibly. Nearly three quarters of respondents also believed AI would reduce the number of jobs available in the United States during the following decade (Marken, 2025).
Edward Luce (2026), writing in the Financial Times, goes further by describing growing skepticism toward AI as an emerging American consensus. What makes his argument particularly interesting is that anxiety about artificial intelligence increasingly crosses traditional political and ideological boundaries. Conservatives, progressives, educators, workers, environmental advocates, and local communities may disagree considerably about why they distrust AI, but their concerns increasingly overlap. The significance is not that Americans have suddenly reached agreement about artificial intelligence, but that skepticism itself has become unusually widespread.
For those of us working in education, the classroom presents another dimension of this problem. Landymore (2026) reports concerns from professors who believe students are increasingly struggling with reading, analysis, synthesis, and independent reasoning after becoming accustomed to delegating intellectual tasks to generative AI. The article also points toward emerging research examining relationships among AI dependence, cognitive engagement, memory, and critical thinking. These concerns should not be dismissed simply because the technology remains relatively new.
A separate report describes an even subtler possibility involving the homogenization of student thought. Students who routinely ask large language models what to think about readings may arrive in classroom discussions with increasingly similar interpretations rather than developing distinctive perspectives of their own. One Yale student described seminar discussions in which classmates appeared to converge around AI-generated interpretations instead of contributing individually developed ideas (Wilkins, 2026). If that pattern becomes widespread, the educational risk is not merely that AI can write student papers, but that students may gradually lose some of the intellectual habits required to have something meaningful to say before the paper is written.
There is an enormous difference between asking an AI system, “Write my analysis of this,” and saying, “Here is my analysis. Challenge my assumptions, identify weaknesses in my reasoning, and show me what I may have overlooked.” Both interactions involve artificial intelligence, but they represent fundamentally different relationships between the learner and the technology. The first potentially delegates the intellectual task itself, while the second requires the human to think before asking the machine to contribute. That distinction should become central to how universities approach AI education.
My concern about cognitive outsourcing does not lead me to conclude that universities should prohibit generative AI. Quite the opposite, because higher education now faces a paradox that will become increasingly difficult to ignore. Universities are understandably concerned about students using AI to avoid learning, while employers increasingly want graduates who know how to use artificial intelligence effectively within their professions. Those expectations are not necessarily contradictory if educators stop treating all AI use as though it represents the same behavior.
Using AI to avoid learning statistics is very different from using AI to examine alternative interpretations of a statistical model after demonstrating that you understand the underlying analysis. Generating an essay about a book you never read is fundamentally different from asking an AI system to compare competing interpretations after you have read and analyzed the book yourself. Likewise, producing Python code you cannot explain is different from using AI to debug code, compare algorithms, document a workflow, or test an approach that you understand. The educational question should therefore evolve from simply asking whether a student used AI to asking what cognitive work the student retained, what work was delegated, and whether the student can demonstrate understanding of the final product.
There are still moments when students need to work without artificial intelligence, particularly while developing foundational knowledge. Someone learning algebra should sometimes solve the equation, someone learning statistics should sometimes calculate and interpret the result, and someone learning programming should sometimes stare at broken code and determine why it does not work. Someone learning to write should also occasionally experience the uncomfortable challenge of confronting a blank page and organizing an argument independently. Productive struggle remains part of the learning process because removing every point of friction may increase short-term efficiency while reducing opportunities for intellectual development.
Foundational instruction, however, is not the endpoint of education. Once students develop sufficient disciplinary understanding, artificial intelligence can become something considerably more valuable than an answer generator. It can function as a simulator, tutor, critic, coding partner, brainstorming partner, research assistant, or even an adversarial collaborator that forces a learner to defend an argument. The objective should not be to remove humans from the intellectual process, but to increase what humans are capable of accomplishing within that process.
This is why universities need a more sophisticated institutional framework than simply labeling AI as either allowed or prohibited. Educational programs should instead begin with foundational competence, where students demonstrate that they can perform essential disciplinary work without relying on AI to substitute for understanding. The next stage should involve guided augmentation, where students learn to document prompts, verify outputs, identify hallucinations, challenge assumptions, and explain exactly what an AI system contributed. The final stage should be disciplinary integration, where students use artificial intelligence within authentic professional workflows resembling the environments they will encounter outside the university.
This progression also means that AI literacy cannot remain limited to generic workshops about prompt engineering. A business student should understand the applications and limitations of AI in business, while a nurse should understand its implications for healthcare and a programmer should understand AI-assisted software development. Teachers need to understand AI-assisted instruction, and researchers need to understand both the opportunities and methodological risks of AI-assisted research. Artificial intelligence literacy must eventually become domain-specific professional literacy if universities expect graduates to enter workplaces where these tools are increasingly embedded in everyday practice.
One of the most important questions surrounding artificial intelligence is therefore not whether these systems will become more capable. They almost certainly will, but the more consequential question for education is what happens to human capability as they do. If artificial intelligence becomes more powerful while people become less willing to reason, question, struggle, verify, and create, technological progress will have produced an educational failure. That outcome, however, is not inevitable.
Artificial intelligence can expose students to perspectives they might not otherwise encounter, provide personalized explanations, and lower barriers to programming, statistics, research, and technical problem solving. It can allow learners to iterate rapidly and receive feedback at a scale that no educational institution could realistically provide through human instructors alone. The determining factor may therefore be less about the technology itself than about the learning architecture surrounding its use. AI should not become the place where thinking ends; it should become a place where better thinking begins.
Perhaps this really is the Semester of Our Discontent. The public is questioning artificial intelligence, communities are questioning its infrastructure, workers are questioning what it means for employment, and educators are questioning what it means for learning. Students themselves may increasingly need to question what constant cognitive outsourcing could mean for their own intellectual development. Those questions should not frighten the AI community because serious criticism can ultimately produce better technology, better policy, and better educational practices.
Those of us who believe strongly in the potential of artificial intelligence have a particular responsibility not to dismiss legitimate criticism. Advocacy without critical examination becomes evangelism, and universities should not be in the evangelism business. Our responsibility is to teach students to understand these systems, question them, challenge them, verify them, and eventually use them with genuine disciplinary expertise. That is substantially harder than either banning ChatGPT or requiring everyone to use it, but it is also much closer to what education has always been supposed to accomplish.
Technology will continue to change, and artificial intelligence will almost certainly become more capable, accessible, and integrated into everyday professional life. Higher education will have to adapt along with it, but adaptation should not mean surrendering the intellectual responsibilities that define education. The challenge is not simply to prepare students to use the most powerful tools available to them, but to ensure they remain capable of thinking independently when those tools are present. Technology changes, but the obligation to teach people how to think does not.
References
Jones, J. M. (2026, May 13). Americans oppose AI data centers in their area. Gallup.
Landymore, F. (2026, March 14). Professors say AI is destroying their students’ ability to think. Futurism.
Luce, E. (2026, August 18). AI phobia is America’s new consensus. Financial Times.
Marken, S. (2025, September 9). Trust in businesses’ use of AI improves slightly. Gallup.
Simonetti, I., & Bobrowsky, M. (2026, August 19). Inside Big Tech’s frantic race to quell the growing backlash to AI. The Wall Street Journal.
Wilkins, J. (2026, April 7). College students losing ability to participate in class discussions due to offloading their thinking to AI. Futurism.
Robert Foreman
Doctoral Candidate, Educational Technology
Central Michigan University
Research Focus: AI-Augmented Exploratory Learning, Cognitive Apprenticeship, and Human-AI Interaction in Professional Skill Development
forem1r@cmich.edu
NhanceData.com
#ArtificialIntelligence #AIinEducation #HigherEducation #GenerativeAI #AILiteracy #EducationalTechnology #CriticalThinking #FutureOfEducation #HumanAIInteraction #EdTech #AIResearch #ResponsibleAI