����ֱ��

����ֱ�� Boulder Professor Mike Klymkowsky uses AI tools to help students develop critical-thinking skills

For many, the idea of artificial intelligence (AI) taking on an expanded role in academia stirs uneasy feelings. Visions of computer-generated tutors and students “writing” essays using a chatbot paint a cold, impersonal destiny for education. However, Mike Klymkowsky, a professor of molecular, cellular and developmental biology at the University of ����ֱ�� Boulder, pictures a different future.

“It’s a tool that students will need to master, but its role will be largely determined by how the institution sets standards,” Klymkowsky says.

Mike Klymkowsky, a ����ֱ�� Boulder professor of molecular, cellular and developmental biology, is using AI tools in the classroom to help students grow critical thinking skills.

Klymkowsky, a veteran educator and innovator, is experimenting with AI not to provide answers to students, but to prompt intelligent questions and facilitate more effective learning. Through self-made AI assessment tools and interactive, personified tutor bots, he encourages students to shift their mindset from memorizing facts to becoming active champions of critical thinking.

Ultimately, Klymkowsky says, the aim of education is to foster skills that extend far beyond the classroom.

“The goal isn’t just to remember the right answer,” he says; “it’s to understand why that answer makes sense and why the other answers don’t.”

Developing more meaningful feedback and assessment

Klymkowsky argues that traditional grading methods, particularly multiple-choice exams, fail to measure true comprehension; they look only for memorization.

Fortunately, he says, AI tools offer a different solution.

“When ChatGPT came out, it became clear to me and everyone else in the universe that these were tools that allowed you to do things you’d always wanted to do,” he explains.

By automating the analysis of students’ responses to open-ended prompts, AI can quickly highlight which concepts cause them to struggle and where instructors can spend more time. Such tasks involving quick analysis of vast datasets to identify patterns are where AI excels, Klymkowsky says.

“Now you can evaluate instructors on whether their learning goals are meaningful and whether the students are achieving them,” Klymkowsky says.

His can reduce days of manually combing through exam responses down to minutes, offering insights that allow him to target lectures more precisely and understand if learning outcomes are being reached.

Klymkowsky says this approach is key to helping students understand not just what they got wrong but why—and how to improve.

From cramming to critical thinking

Klymkowsky’s approach to AI addresses a long-standing challenge in academia: the prevalence of rote memorization.

“Whenever a class starts using multiple-choice questions to answer, forget critical thinking. You’re not asking them how they got the answer; you’re asking them whether they recognize it,” says Mike Klymkowsky. 

“Whenever a class starts using multiple-choice questions to answer, forget critical thinking,” he explains. “You’re not asking them how they got the answer; you’re asking them whether they recognize it.”

Klymkowsky, an avid proponent of exploratory, inquiry-based learning, created an AI “tutor bot” called “Rita” to enhance his students’ learning. The bot uses a technology known as retrieval-generated augmentation and is trained on information provided by Klymkowsky, including lecture materials and textbooks. Limiting the bot’s knowledge to a select dataset prevents it from “hallucinating”—making up potentially incorrect or misleading answers to questions it doesn’t know.

“Our bots, when you ask them a question they don’t know, they say ‘I don’t know.’ If you ask ChatGPT or Claude a question, it’ll answer whether it knows it or not,” he says.

Klymkowsky views Rita as a patient guide capable of leading students through complex materials at their own pace.

“These bots don’t just spit out answers,” he says. “They respond based on what students already know and ask follow-up questions to deepen their understanding.”

He also explains that the bots can be tailored to specific disciplines with a custom knowledge base. Keeping the bots within their trained parameters ensures students can rely on them to deliver accurate information without straying into unfamiliar territory.

“You want to have the bot be focused on what the learning outcomes of the department are,” Klymkowsky says. “So, if students are engaging with a bot in a biology course, that bot is designed to know what it knows and what it doesn’t know.”

Tutor bots like Rita use the Socratic teaching model to promote critical thinking. They work with students to challenge their assumptions and develop solid explanations for their reasoning.

“Imagine being able to practice asking questions with a bot that makes you feel appreciated because it never loses its patience, right? It’s never snarky,” Klymkowsky says.

Rita won’t simply ask a student for the answer. In the form of a conversation, the bot asks for a reflection on why the student believes their answer is correct—or why it isn’t—to help them grasp the underlying principles of a given topic.

“The goal is not to memorize facts, but to understand the why and how behind them,” Klymkowsky says. “It’s about cultivating the kind of thinking that lets students ask the right questions—and teaches them how to start finding answers independently.”

Engaging students beyond the classroom

In addition to Socratic tutor bots, Klymkowsky is using Notebook LM to as a novel tool to spark curiosity. As with Rita, he creates these podcasts using a limited dataset, such as a course textbook.

The AI tool then turns the input into a two-way conversation between virtual speakers. Despite the surreal experience of listening to an entirely non-human conversation, the format allows students to explore high-level information in a more accessible style through a medium many younger adults favor.

“The goal with these podcasts is to give students a jumping-off point—something that piques their interest and motivates them to dig deeper,” Klymkowsky explains.

Each podcast episode introduces a biology concept, immersing students through storytelling and examples.

While the application is promising, Klymkowsky knows producing such content is a tricky balancing act of depth and attention span: “What is the attention span of the student? How long are you going to keep them on task before you ask them to do something themselves?”

Despite this challenge, Klymkowsky believes AI podcasts can complement classroom learning by acting as conversation starters.

“It’s more about using the podcast to motivate students to go read the book or the chapter—or to ask questions that they wouldn’t otherwise consider,” he says.

From there, students can bring their questions into class discussions or interact with a tutor bot to reinforce their learning.

By embracing AI tools like Socratic tutor bots and podcasts, Klymkowsky believes it’s possible to create an educational space where students can deepen their understanding through diverse content formats while cultivating a habit of lifelong learning that goes beyond a multiple-choice bubble.

Fueling curiosity, one question at a time

As technology continues to shape academia, Klymkowsky emphasizes that AI, when thoughtfully applied, needn’t be the villain. Instead, it can be a powerful catalyst for cultivating critical thinking.

“If you don’t understand a thing, can you ask an intelligent question?” Klymkowsky says.

With AI as a partner, he says he believes students can learn to ask those questions, and that AI can be used to develop curiosity and intellectual resilience—skills that will serve students far longer than a perfectly memorized breakdown of the Krebs cycle. 

Did you enjoy this article?  Passionate about molecular, cellular and developmental biology? Show your support.