Editor's Note: Smith College students in a Journalism in the Field course were asked to practice reporting in the community. Students then worked with their professor Dusty Christensen and NEPM to to edit and promote the pieces. This is Rowan's piece. Read the others here.
In the post-COVID era, most know the ritual by heart. Muted microphones and blank squares. The angling of the computer screen, the half-second video delay, the internal calculus for when to unmute. The dreaded Zoom room, where conversation flattens, eye contact is a shot in the dark and every silence feels just a second too long.
But this particular Zoom meeting feels different. That’s because it carries the weight of possible discipline. If you’re a first-year, there’s a measure of grace, according to a faculty member and a student who have attended these meetings. You’ll most likely get off with a warning. If you’re older, the consequences can be more severe. In the Zoom grid, faces and names settle into place, four students and three faculty, as they wait for the last participant to join.
“I wish I hadn’t” is an admission that the Smith College Academic Integrity Board hears often. And nowadays, Tina Wildhagen, the faculty chair of the board, says the vast majority of students are admitting to the same transgression: using generative AI on their schoolwork. Most people admit to it right away, she says. For many, the use of the technology is less about cheating than it is about pressure, she added. It’s not that students are drawn to using generative AI because they are lazy or want to cut corners. It’s “because they feel desperate,” said Wildhagen, a sociology professor at the college.
Today, experts say, technology has changed both undergraduates' academic expectations and their approach.
The emergence of generative AI began suddenly in 2022, when ChatGPT entered the public's view. Since then, AI tools have proliferated across college campuses. Students have access to task-specific AI assistants for everything from coding to writing papers and even "humanizer" programs that mask the use of AI altogether.
This issue has captured the attention of professors and college administrators worldwide, including in western Massachusetts, where last month the Five College Consortium held an event exploring the impacts of AI on higher education.
Katherine Douglas, the interim vice president of academic and student affairs at Holyoke Community College, says that people have long been interested in using instruments that make tasks faster or more efficient. She gave the example of how artists in Paris in the mid-19th century were used to being commissioned for portraits and landscapes until a new technology came in and disrupted their industry. It was called photography.
All of a sudden, people could get their portraits and landscapes faster and cheaper. As a result, the painters had to adapt. They needed a new style, something distinctly different from what the camera offered. This is how we arrived at impressionist art, she said.
Douglas says that this story parallels what she and her colleagues are grappling with right now.
“You have a traditional way of doing things, and then technology totally upends it. And what do you do with that? That’s the challenge I see for higher education with AI.”
For some institutions, the emergence of AI raises questions about academic integrity, cheating and the future of students’ intellectual labor. Others, however, are embracing the technology as the future. Springfield College is launching a data science program this fall. In a promotional video, James O’Brien, the department chair, says that “everyone is using it,” and that the college’s new program will train students to “not just use AI,” but to “become and create the AI.”
A Higher Education Policy Institute survey found that AI use among undergraduates rose from 66% to 92% in the past year — a trend emerging at local schools, too. At Smith College a spring 2025 survey found that, from a pool of 737 respondents, 64% reported using generative AI at least once for tasks categorized as an “academic assistant,” while 51% reported using generative AI at least once as a “thinking replacement.”
“You have a traditional way of doing things, and then technology totally upends it. And what do you do with that? That’s the challenge I see for higher education with AI.”— Katherine Douglas, interim vice president of academic and student affairs, HCC
Some higher-ed leaders are excited about the opportunities AI presents, like Douglas, who highlighted the technology's capacity to speak in a student’s first language – something especially useful for international and first-generation students. On the other hand, others have reservations about its broader impact, from the steep environmental costs of AI data centers to algorithmic bias in the data sets used to train AI.
Kye Barker is an assistant professor of government at Smith College. He said he tries to limit students’ temptation to use generative AI, but feels he is creating half-measures against the power of Big Tech, which could , with the new tool, “destroy higher education, whether or not that’s their intention.”
When Wildhagen asks students who come before the Academic Integrity Board at Smith College why they felt the need to use generative AI when explicitly told not to, she says they often turn the question back on her, asking how she thinks they got to Smith College in the first place: “I got here by making no mistakes.”
That mindset, she explains, has been shaped by a growing weight of the 4.0-to-post-grad-job pipeline and an increasingly uncertain future. Students feel trapped in a pressure cooker where anything below an A is a failure, and generative AI seems like the only way to keep up. The result, she worries, is a quiet erosion of error and failure as pathways to intellectual growth.
A history of AI and a history of cheating
In 1950, the renowned British computer scientist Alan Turing published a paper called “Computing Machinery and Intelligence,” questioning whether machines can think. The Turing Test, as it became widely known, became a landmark thought experiment — a test of a machine’s ability to exhibit intelligent behavior that is indistinguishable from that of a human.
Today, AI models are passing the Turing Test more than real humans. In a recent three-party version of a Turing Test, ChatGPT-4.5 was mistaken for a human 73% of the time when assuming a human persona, meaning that newer versions of AI can convincingly mimic human behavior.
The field of AI has fluctuated over the years, but by the late 1990s and early 2000s, researchers began developing neural networks — computational models inspired by the human brain. Given enough data to train on, these networks could learn to recognize patterns, make predictions and improve their performance over time.
This is what current chatbots are doing on a larger scale: training on massive amounts of data to perform complex reasoning, multi-step tasks, and multimodal inputs based on probability and context.
So, while AI tools look different today than they did decades ago, AI itself is not new. Even in the past decade, Amazon had Alexa, Apple had Siri, and Google had Google Assistant before chatbots debuted.
Yet, while the AI used in programs like Amazon's Alexa responds to commands, generative AI creates new content from learned patterns. Since the public release of ChatGPT in November 2022, generative AI has been integrated into both industry and academia at a remarkable speed. OpenAI’s ChatGPT now has 900 million active weekly users, according to the company. As technology evolves, so do work methods.
Some, though, are wary of the hype.
Casey Bohlen, a history professor at Smith College, remembers how, about 15 years ago, as a graduate student at Harvard University, massive open online courses were pushed into his curriculum. He said that for-profit tech companies like Coursera and Udacity convinced institutions — some of whom invested colossal sums of money on the technology — that the future of higher education was going to be online. Courses would be live-streamed, and there would be digital seats for sale globally.
But MOOCs, as they became known, did not become the future of education. Bohlen says there are “a lot of instances of supposed innovations that just never catch on, or are bubbles that collapse.” This is why he maintains a default position of skepticism about whether new technologies like generative AI will be the future. He says he needs to see some evidence of the impact of AI on student learning before he’s ready to implement it into his classrooms.
Breakdown of trust in the classroom
Bohlen said, what is most frustrating about AI for him in a classroom setting is how it has “fractured trust at a deep level.”
The fracture starts small. Bohlen sits in his office, grading papers, comparing two that seem eerily similar. Across campus in a dorm room, a student hovers over the submit button, wondering if her words are “good enough” to stand on their own.
Professors interviewed for this article say they are concerned that their students are cheating with generative AI. But, as Nikko Stevens, a professor of statistical and data sciences at Smith College, points out, students have always cheated.
“I was once a student,” they said. “We just cheated differently.”
What matters to Bohlen, he said, is persuading students that doing the work themselves is more worthwhile, and that the process of learning has value beyond a grade.
But this is really difficult for professors like Bohlen in the age of generative AI. He said that when he receives a great paper, “there’s this little part of me that feels like, ‘Did they write this?’”
But, “I don’t want to be a cop,” he said. “The costs of being wrong are really bad.” Bohlen said that reporting a student degrades trust in its own way: undermining the collaborative environment on which learning depends. This is why, he said, he has never reported a student for suspected AI use.
If forced to choose between preserving the trust central to a productive classroom and policing the use of this new technology, many professors say they are choosing trust.
Tina Wildhagen, the faculty chair of Smith’s Academic Integrity Board, said most students accused admit their wrongdoing right away. But the breakdown of trust appears to be growing among both faculty and students.
And in some cases, students are fighting back against accusations that they misused AI. Some have been able to overturn their discipline, like a student at Long Island-based Adelphi University who won his lawsuit earlier this year. Others have failed.
Parents of a Massachusetts high school student sued Hingham High School after the school concluded in December 2023 that the student had cheated on an AP U.S. History assignment by copying and pasting generated AI text, then turning it in without attribution. A federal judge ruled that Hingham High School’s punishment for using AI to complete an assignment could stand.
It’s unclear just how many students are being caught and disciplined for the use of generative AI in their classes. In response to a public records request for disciplinary data, the University of Massachusetts Amherst — the state’s largest public university — said that it doesn’t yet keep data on the subject. The university codes unauthorized use of AI as "cheating" along with other forms of cheating, a spokesperson said.
The integration question
Nikko Stevens, the professor of statistical and data sciences at Smith College, recently attended a faculty meeting in which the speaker was tasked with explaining the fundamentals of AI. When questions were invited, Stevens said the faculty cohort erupted. Issues of data centers, environmental impact, cognitive overload and algorithmic racism all came bubbling to the surface. What stood out the most to Stevens was how emotional the conversation felt.
“Are we worried about our jobs? Are we worried about our students’ jobs? What is our affective relationship to what is happening?” Stevens asked.
This tangled web of concern points to one specific question many faculty are grappling with: Should they integrate AI into their curricula?
Alli Martel, the digital technologies librarian at Springfield College, says that next fall, the college will offer an AI data analytics program. Martel says that higher education institutions should be setting their students up to do the best work that they can.
So, she says, “it would be useful to have some AI guidelines,” and admits that what exactly these guidelines will look like is currently being discussed at Springfield College. She says it is difficult when there isn’t much institutional guidance for AI use. But, she said, “I understand why.” AI is evolving so rapidly that “it’s difficult to keep up.”
Last August, Katherine Douglas, the Holyoke Community College interim vice president for academic and student affairs, asked her faculty about their campus concerns.
The response? “We’re drowning with AI. It’s everywhere.”
Douglas said she answered the call for help. This fall, she said members of HCC’s faculty council conducted studies and focus groups to identify their AI-related concerns. From there, Douglas proposed a three-element approach to AI.
The first element is a faculty sabbatical this fall that will send a faculty member to research AI in a writing-intensive classroom. The second is an AI task force, which has a full-time faculty member from each academic division, an adjunct faculty member from each department, a representative from disability services and someone from the registrar's office.
The third element is building what Douglas calls an AI Academy, where faculty will volunteer to investigate how they want to integrate AI into their existing coursework. But, she says, coming up with policy is something she wants to “be careful about.” Douglas aims to make recommendations for guidelines, rather than creating policy, because she says the landscape of AI is changing so fast that “whatever policy we put in place is going to be outdated by the time we would be able to implement it.”
Her overall goal is to preserve the human connection that she believes is at the heart of learning, while also crafting an ecosystem that “accepts the reality that AI is here.”
She fears that if “we as educators don’t truly, authentically develop the critical thinking in our students, that’s an incredible disservice to the individual and to society.”
Luca Capogna, a math professor at Smith College, uses AI in his classroom as “an additional layer, not a substitute.” Just this academic year, Smith began piloting a platform called Smith Campus GenAI, developed in partnership with UMass Amherst. The platform gives students access to major large-language models and allows professors to build AI agents — like virtual TAs, writing assistants or role-play simulators — for learning and feedback.
Capogna has already integrated the tool into his math classes and says he would “encourage other professors to use it as a possible tool.” In general, he says, even people who do not work in fields that directly relate to machine learning can “equally benefit from AI,” because it gives access to a broader body of knowledge beyond the limits of individual humans.
“The critical thinking part in how to put together these pieces is something that we still have to do,” he said. But AI is not going away anytime soon. “It’s going to be more integrated in society, and if we isolate our students from it, we are doing a disservice to them because we are not providing them with what is being called for in this moment.”
Kye Barker is an assistant professor of government at Smith. He calls generative AI “a large-scale assault on the lived value of what we’re trying to get students to take out of these four brief years.” He says that when students rely on generative AI to brainstorm, research and struggle through their work, they aren’t engaging in key processes that build the critical thinking skills college is meant to develop.
Barker also warns that higher education, which he calls “an extremely fragile experiment in American life,” is being commodified by companies like OpenAI and Anthropic, which profit from students’ use of their products. Because generative AI can produce student-level work at no cost, Barker fears that it could “undermine the economic value of a degree,” leading employers to question what students have actually done themselves.
The Architecture of Perfection: Student voices
Talk to students on college campuses today and many feel like Smith College student Grace Macedonia: anxious as they look at the job market.
“My field is impossible to get a job in right now,” the art history and Italian double major explained.
Shriya Gautam, a senior computer science and statistics double major at UMass Amherst, said this is one of the reasons students there founded MassAI, an organization that helps teach their peers how to use AI. Guatam says that she feels the pressure “from the industry side for computer science students to get really good at using AI and then also be building AI or at least having basic AI literacy.”
[AI] is going to be more integrated in society, and if we isolate our students from it, we are doing a disservice to them because we are not providing them with what is being called for in this moment. — Luca Capogna, Smith College professor
Guatam says that a lot of UMass students “do use AI on homework or even preparing for exams.” But, she says, she doesn’t think that using AI alone will get someone through a degree. She said students are still “putting in effort to actually learn.”
While many professors frame AI use as misconduct, some students frame it as workflow integration.
Guatam was in a graduate-level, physics-heavy computer-science class. She had to write code for the project, but said she had “no idea how to get from point A to point B.” She tried looking online, but it was too complex to decipher. She felt that she “kind of had to go to AI.”
But she said that she asked the AI chatbot just basic questions. Once she understood the material, she finished the project on her own.
At the same time, Guatam says that the pressure to use AI also comes from the reality of college nowadays. “If you’re not doing well in this class, you need to make sure you’re doing everything you can to do better, and that includes using every single tool at your disposal, including using AI.”
Ceshia Hegarty is a recent graduate of the soon-to-close Hampshire College, which is known for its narrative evaluations instead of grades — a system some praised as a way to encourage intellectual growth and risk-taking in the AI age. She said she used AI for the first time in her entire college career this spring. She had never taken a statistics course before, but needed to interpret statistical graphs for her psychology thesis.
“Could I call a friend? Yeah. But was it easier to plug it into ChatGPT? Yeah.” However, she said she doesn’t trust it and was “a little scared of it.”
But while some — including Bohlen, the Smith College professor — view systems like Hampshire’s as a better path forward for student success, it’s not clear that higher education is moving in that direction. In April, Hampshire announced it would be shutting down at the end of the year due to financial difficulties.
The Future: AI in Higher Ed?
A student at Smith College walked into the yellow light of the Spinelli Center for Quantitative Learning in the fall of 2025, struggling through an intro-level economics course. She came often for tutoring, said Catherine McCune, the center's director.
This semester, the student returned with different news. She earned the highest grade on the most recent exam out of her whole class. The shift, according to McCune, wasn’t about getting the right answers all the time, but about learning how to think.
As McCune and her colleagues have observed, while human tutors scaffold study skills as well as content, “AI may be focusing on the content rather than helping you learn.”
The Spinelli Center is increasingly finding that students who use AI on take-home assignments cannot retain class content when putting their understanding to the test, McCune said. In a current Smith College intermediate economics tutoring group, tutors reported to McCune that many of their economics students “are getting in the 90s for their problem sets, but failing the exams.”
This is the difference between using AI and using a human tutor, she said. While ChatGPT “will give you the answer,” all of the tutors at Spinelli are “trained not to give you the answer, but to nudge you in the right direction.”
As professors and administrators look ahead, many express a shift away from questioning whether students use AI and toward ensuring students’ learning is deepened rather than replicated or replaced.
Alli Martel, the digital technologies librarian at Springfield College, calls herself AI agnostic. “It’s here,” she said, but she has reservations.
“If you’re using a calculator all the time, then you might not remember how to do long division.” She says that just like a calculator, AI is a tool. And, she says, any tool can be used for good or bad.
Martel says that she is curious to see how many students will have been so reliant on tools like ChatGPT in high school that they struggle with the transition into college. Because, as Martel points out, “professors will absolutely know that you are using ChatGPT to write the paper. And they will fail you.”
But, given the current environment students are graduating into, with economic uncertainty and a scarce post-graduation job market, students say the pressure to rise to the top has never been higher. AI offers an escape valve.
And that’s now, before some predict that generative AI will make many entry-level jobs obsolete.
With stakes that feel monumentally high, some question who wouldn’t feel compelled to use AI. That’s how Gautam, the UMass Amherst student, feels.
“If you’re not doing well in this class, you need to make sure you’re doing everything you can to do better, and that includes using every single tool at your disposal,” she said. “Including using AI.”
