Education is in the throes of an AI reckoning. In a little more than three years, artificial intelligence has morphed from a science-fiction cautionary tale of robots ruling the human race into a very real debate being argued over in lecture halls and conference rooms across the nation. 

Attempting to wrap their heads and hands around a transformative technology that is, at best, the Wild West and, at worst, the end of human intellect as we know it, educators are at odds. Is AI a crutch that threatens student brainpower, or is it a powerful tool that can enhance their learning? Is there a right way to use AI in the classroom, or should it be banned outright?

At the University of Houston, educators are in the early stages of exploring the positive and negative boundaries of AI, while enthusiastically encouraging its use over prohibiting it. It’s a conscious decision that puts UH at the forefront of a movement that could fundamentally alter the way students learn.

UH professors Amy Novak, Susie Gronseth and Minwoo Lee are all early adopters of AI in their classrooms, viewing the range of tech tools as exciting and efficient supplemental learning aids students can use to automate tasks, enhance reading comprehension, improve critical thinking and more. Their consensus is that we shouldn’t run from AI nor hand over the keys but rather teach students how to drive it.

“The train has left the station,” says Novak, a senior professor of practice of marketing who teaches professional selling, personal branding and business communications. “We gotta make sure our kiddos are on it.”

Brains First, Bots Second

Welcoming AI into the classroom, however, doesn’t come without its own kind of chaos. Universities struggle with prevalent cheating and plagiarism, as well as the dilution of students’ interest, patience and effort in exploring subjects without the aid of an artificially intelligent sidekick that can spit out a thousand words in less than a minute. On the other side of the lectern, AI has given rise to false detectors, causing educators to inaccurately accuse students of intellectual infringement.

Luckily, Novak, Gronseth and Lee are aware of the issues and stand at the ready to solve them. In their view, there is a right way to use AI in the classroom, and though their reasoning differs slightly, they all agree on one fundamental rule: Use your brain first. Use AI as a supplement.

Gronseth, a clinical professor in the Learning, Design and Technology program area of the College of Education, takes the philosophy one step further, stressing that AI should be used with clearly set goals, keeping outcomes front and center. She says AI should be tapped only when it meaningfully supports those goals and not for just the trendiness of it.

“Students have to go beyond what AI puts out,” she says. “We think more critically, so challenge the output and use other sources to compare. AI can be a thought partner to help you get started.”

Lee, an associate professor and director of the Hospitality Analytics and Innovation Lab at the Conrad N. Hilton College of Global Hospitality Leadership, first primes his students with AI literacy, then teaches them to distinguish between primary and supplementary tasks.

“If I ask you to write an essay, the primary task is writing the essay based on your experience,” he says. “If you use AI for that primary task, then you are using AI as a crutch. But if you use it for proofreading or revising for a specific audience, you are using it in the right way: as a support mechanism.”

But oh, how easy it is to just copy and paste that initial output into a Google Doc, right? Well, not so fast. A common misconception among students who use AI is that the information they get from language learning models is properly sourced and factual. Sure, some of it is. But sometimes AI outputs junk — or, to use the industry’s technical term, hallucinations.

LLMs prioritize likely word sequences over actual facts, offering prompters responses that resemble human speech, sprinkled with “facts” and “quotes” that sound confident but carry no weight in truth. Lee knows this all too well.

“Sometimes I check AI with my research,” he says. “I asked it to find five areas of my research focus, and the results were really funny. The first four were correct, but the fifth pertained to golf. I have never written a paper about golf. But because there is a professional golfer named Min Woo Lee who won a Houston golf tournament, AI pulled in that information.”

If professors aren’t immune to hallucinations, students certainly won’t be. So when students get it wrong, where’s the line, and how should educators teach them not to go near it?

“We have to enable students with AI, but they have to think, be curious, and maybe even a little dubious,” Novak says. “But if you get it wrong, it’s on you.”

Guardrails, including tools such as Grammarly’s AI Detector and Turnitin, which flag heavily AI-written materials, are used by UH instructors. But even those are flawed, spitting out unreliable and inconsistent results. To avoid getting it wrong herself, Gronseth relies on spotting voice shifts in a student’s work to alert her to AI misuse. She also mentions transparency statements from students.

“Our courses encourage our students to use AI in certain parts, but we require a statement as part of the submission that describes how they used it,” Gronseth says.

AI as a Classroom Multiplier

Despite its many concerns, UH is eager to fully deploy AI in its classrooms, as the benefits of using the tech as a learning tool far outweigh the negatives. In Novak’s Professional Selling course, which teaches the consultative sales process, she implemented AI role-plays in the spring of 2024 to 1,350 students for the semester-long sales project. Its impact cannot be denied. As a core curriculum course for the C. T. Bauer College of Business, the teacher-to-student ratio threatens to exceed the limit of one professor’s resources. That is, before AI tool Second Nature.

“We simulate the entire semester sales project on Second Nature,” Novak says. “Students use it for role-play, to perfect their presentations, and to generate prospect emails and initial outreach, tweaking each response with edits and then submitting their final project.” She now applies the same technology in her Personal Branding class, which implements mock interviewing to teach students how to sell themselves.

“I used to do mock interviews. It took three class periods,” Novak says. “Now if a student wants to do it 10 times over, they can. In fact, I had one student, a non-native English-speaking student, who scored a 26 on their first encounter with the software, and then ended up with a 100 after doing it 39 times.”

In the College of Education, the story rings a similar bell. An expert in Universal Design for Learning, an educational framework designed to “improve and optimize teaching and learning based on what we know about the human brain,” Gronseth uses AI to support deeper understanding and different learning styles. She mentions AI’s ability to break down complex text into audio or visuals, provide vocabulary explanations and translate texts into multiple languages.

“The students were taking the AI translation and comparing it to other forms of translation of a particular passage,” she says of an AI translation session held at UH. “It was so interesting to see, because I’m not a translator, but I never thought about it as being this art form. It’s not like there’s one translation for the passage, because there are artistic decisions that the writer makes in the way that they craft their text, so it really makes you appreciate that text more.”

Beyond complex reading and translations, Gronseth views AI as a tool for enhancing metacognition. She experiments with programs like Stanford’s Ask.Smile, which prompts students to ask better questions. The user begins by asking a question, and AI rates the quality of the query based on Bloom’s Taxonomy, then coaches the prompter to ask more complex questions.

“It’s not a standalone tool,” Gronseth says. “It comes in combination with an instructor that’s creating the experience and debriefing at the end.”

Over at the Conrad N. Hilton College of Global Hospitality Leadership, AI is beginning to play a significant role, particularly in Lee’s Hospitality Analytics and Innovation Lab, where he develops education modules for data analytics and social media marketing. Lee empowers the use of Microsoft Copilot, an AI companion that promises to supercharge productivity and protect data privacy, but he insists students use the tool only after the primary assignment has been completed.

Lee also uses AI for hands-on learning with communication and writing, citing AI’s ability to convert one piece of content into multiple audience-specific versions (think LinkedIn captions versus an executive report). One of the most promising uses of generative technology, in his view, includes the ability of AI to analyze hotel reviews, populating word clouds and surface patterns hospitality employees can use to then handle complaints and address problem areas.

“If you have those online reviews, you can simply analyze those reviews to create a word cloud and find the service failures,” he says.

His work on innovating hospitality and the tourism industry, described in “AI in Hospitality and Tourism: Rewriting the Customer Journey,” a case study that earned him and his international research team the Best Presentation Award at the International Society of Travel and Tourism Educators’ 44th annual conference, explores AI’s role in “enhancing hyper-personalization, operational efficiency and stakeholder value, while addressing challenges like ethical concerns and workforce impacts.”

Bright Horizons — With Boundaries 

As the University moves forward with AI use in the classroom — students, faculty and staff now have unlimited access to data-protected Gemini 2.5 and daily prompt-limited Gemini 3.0 — further concerns over originality and ownership are warranted, as are worries about data privacy. Where is all this stuff living, anyway? But with the proper guardrails in place and a University-level committee dedicated to promoting AI transparency and balancing innovation with integrity in AI adoption, the future looks mighty bright illuminated by AI.

“It is not the strongest that survives,” Novak says. “It is the one that is most adaptable. If we don’t embrace AI and teach around the good, ethical and actual ways, then we’re doing a disservice to the next generation.”