Keeping critical thinking at the core of AI in higher education
Generative artificial intelligence is quickly changing the way students learn, but it should not make human judgment obsolete. If anything, the growth of generative AI makes teaching students how to question, evaluate and reason more important.
There is understandable wariness surrounding the use of generative AI in higher education. Students can prompt an AI tool to quickly summarize a lengthy reading, solve a complicated equation or draft an essay. This leaves faculty wondering what happens to the learning that used to occur in the process of students manually doing those things. And there might be a larger question as well: If technology can increasingly do the work, what exactly should students be learning to do?
The answer may lie less in protecting tasks students have traditionally completed and more in strengthening the critical thinking skills needed to ethically apply the technology.
Preserve critical thinking
The challenge for higher education leaders is not to preserve each task and workflow process as it existed prior to generative AI. It is to make sure the thinking skills those tasks were intended to develop remain central to the educational experience.
That distinction matters.
A student who asks an AI tool to generate an answer is not necessarily learning less. A student who blindly accepts that answer without question, however, may be. AI tools that make information easier to access can also make it easier to skip the intellectual work of deciding what is credible, relevant or incorrect. This is where critical thinking is crucial.
Tom Zhang, a University of Florida lecturer in stage management, has approached that challenge by showing students what AI can do and where it can go wrong.
“To monitor it well, we have to know the basics of how AI works and the basics of our discipline,” Zhang said. "Demonstrating the critical thinking process of AI, how it can mess up and how to exercise caution when using it, this turns AI from something students simply use into something they must learn how to evaluate.”
At the University of Florida, the AI across the Curriculum initiative reflects a similar distinction. Rather than putting AI in opposition to learning, the initiative helps students understand how these tools fit into the disciplines they are studying and the professions they will eventually enter. And as the workplace increasingly uses AI, students need opportunities to understand its capabilities, limitations and ethical implications.
Andrea Ramirez-Salgado, an instructional assistant professor in UF’s Department of Engineering Education, argues that students need to understand AI beyond the most familiar generative tools.
“Demystify the idea that AI is just ChatGPT and go further with AI concepts and foundations,” Ramirez-Salgado said. “Go into how AI is trained and what the implications are.”
That foundational knowledge matters because AI literacy is not static. Students enter classrooms with different assumptions about AI tools, and those assumptions can shape how they use them. Ramirez-Salgado believes understanding students’ starting points is increasingly important because “AI literacy is evolving as fast as AI is evolving.”
AI in higher education
Historically, educational institutions have asked students to demonstrate what they know by producing some form of work: a paper, a presentation or by showing the steps of their work when solving an equation. But when generative AI can increasingly produce a worthy first draft, the product alone tells us less about how a student arrived at this information.
This leads educators to new questions. Can students explain the reasonings behind their answers? Can they identify weaknesses within their arguments? Can they compare competing ideas, defend a decision and explain why they have changed their minds?
These skills do not become irrelevant when generative AI improves. They become the skills that help determine whether AI tools are useful to students.
Zhang has seen that shift play out in his own courses. As AI has made some technical and procedural tasks easier to automate, he has put greater emphasis on transferable skills such as critical thinking and collaboration.
“These transferable skills, including working with others and critical thinking, are a really important part of what we’re teaching,” Zhang said.
Darius Brown, a senior lecturer at UF’s Digital World’s Institute, has reached a similar conclusion from the perspective of digital arts. For Brown, the question is not whether AI will be a part of creative work. It is whether students can use it while continuing to develop the human skills their fields require.
"Communication and teamwork and problem solving are the big three," Brown said. "And if we are using AI to help you get better at those core competency skills. You are the art director."
That perspective also changes the conversation around whether students should be allowed to use AI. There are legitimate concerns about overreliance on AI but classifying AI as forbidden or off-limits can create another problem. Graduates will enter workplaces where these tools are increasingly commonplace. As such, they need to know not only how to use them, but when to question them, when to verify them and when to ignore them.
Neither extreme serves students particularly well. The better solution is to embed AI literacy grounded in human judgment.
Pairing human ethics with AI tools
Integrating AI literacy begins with teaching students to approach AI-generated responses with the same intellectual curiosity they should bring to any other source.
Ramirez-Salgado argues that AI should not be treated as a new learning objective simply because it is a new technology. Instead, it should support the skills students are already expected to develop.
That distinction can help faculty avoid one of the most common pitfalls of AI integration: using a new technology simply because it is available.
Before introducing an AI tool, faculty should ask whether it meaningfully contributes to student learning or is simply being added because it is trending. Ramirez-Salgado also said the first question should be whether the technology will contribute to students’ learning. The same principle applies to assessment. If AI can complete an assignment with minimal student involvement, educators may need to reconsider whether the assignment is measuring the skill they intend to assess. This means educators should make the thinking behind the work more visible.
This may come in the form of faculty asking students to critique an AI-generated response, identify any errors, compare AI responses with their primary source, explain why they rejected a suggestion or document how their thinking changed during a project.
In his own course design, Brown emphasizes transparency around why AI is used and how it connects to specific learning objectives. He even builds courses where some modules deliberately do not use AI because the technology does not help advance the goals of that particular lesson. The point is to use AI with intent.
This means giving students room to engage with AI differently. For universities, that requires more than asking individual faculty members to figure it out. They need to support professional development, opportunities to experiment and clear expectations for ethical AI use.
Moving forward with AI integration
AI is constantly evolving, and effective teaching practices will need to adapt to keep up. Ramirez-Salgado noted that instructors must be willing to navigate uncertainty when bringing new technologies into the classroom because questions and challenges will inevitably arise.
For policymakers and higher education leaders, this may be one of the most important parts of the conversation. The value of a college education is not simply whether graduates can use the latest technology. Technology will change. What endures is their ability to adapt alongside it.
AI can help students build those skills, but it can also make it easier to skip the learning required to develop them. Higher education’s challenge is to ensure students learn not just how to use AI, but how to think alongside it.
Moving forward with AI integration
AI is constantly evolving, and effective teaching practices will need to adapt to keep up. Ramirez-Salgado noted that instructors must be willing to navigate uncertainty when bringing new technologies into the classroom because questions and challenges will inevitably arise.
For policymakers and higher education leaders, this may be one of the most important parts of the conversation. The value of a college education is not simply whether graduates can use the latest technology. Technology will change. What endures is their ability to adapt alongside it.
AI can help students build those skills, but it can also make it easier to skip the learning required to develop them. Higher education’s challenge is to ensure students learn not just how to use AI, but how to think alongside it.
This content was provided by the University of Florida. The editorial staff of The Chronicle had no role in its preparation. Find out more about paid content.
Built with Shorthand