Exploring How AI Personas Can Boost STEM Skills: Insights from North-West University
STEM education is at the core of Mindjoy’s mission, and we’re always looking for new ways to understand how AI can support learning in the most effective and efficient ways possible.
Recently, a team of researchers from North-West University (NWU) in South Africa - Liezl van Dyk, Roelof Burger, and Jacoba H. Bührmann - published a study that dives into exactly this.
Their paper, “Deploying Multidisciplinary AI-Personas to Enhance Critical Thinking and Communication Skills in STEM Education”, explores how different AI personas can help students improve critical thinking, decision-making, and communication skills in complex, real-world scenarios.
Why This Study Matters
STEM students, especially in engineering and science, often need to work with people from different disciplines, whether that’s finance, environmental science, policy, or manufacturing, to name a few.
Success is about technical know-how as well as being able to communicate clearly, reasonably weigh up multiple perspectives, and make logical, informed decisions.
With this in mind, the NWU research team wanted to see if AI personas - virtual characters representing different roles - could effectively simulate this kind of multidisciplinary teamwork, helping students develop these essential skills before entering professional environments. The idea was to test out if the personas were authentic enough to be useful and if they were effective in providing students with “practice” for real world scenarios.
Two Case Studies: Decision Science and Environmental Science
The researchers ran two different case studies.
Environmental Science (Natural and Agricultural Science students)
Students wrote reflective essays on environmental worldviews. They interacted with AI personas that represented different ecological perspectives, each of which encouraged them to compare viewpoints, reflect critically, and practice articulating their own conclusions.
Decision Science (Industrial Engineering students)
Students were tasked with deciding the best location for a bakery acquisition. To do this, they had to evaluate multiple options using a method called the Analytic Hierarchy Process (AHP), which involves comparing options and weighting different decision criteria.
For the sake of the research, AI personas acted as the CEO, financial manager, marketing analyst, and distribution manager, providing information and expertise for each area. Students then had to gather information, identify inconsistencies, and make reasoned recommendations.
How Students Interacted with the AI Personas
In both case studies, students engaged directly with the AI personas to simulate real-world conversations. The personas were designed to act like busy professionals, offering concise, role-specific insights rather than teaching technical concepts. Essentially, behaving more like actual people would in a real-life scenario.
In the Decision Science case, students even had to identify and resolve inconsistencies in the AI responses, mimicking the way decisions are checked and double-checked in professional settings.
What Did the Study Find?
The study showed that AI personas can:
- Encourage students to think critically and communicate clearly
- Simulate realistic multidisciplinary collaboration
- Help students practice decision-making and reflecting on multiple different (sometimes conflicting) perspectives
The researchers emphasized that AI personas work best when paired with structured guidance and instructor support. Essentially, it echoed a sentiment we’re familiar with - AI doesn’t replace teaching; it offers a dynamic, interactive supplement.
Why We Find This Exciting
What makes this research particularly compelling is the way it opens up new possibilities for simulation as a learning and assessment method.
Traditionally, assessing skills like critical thinking, decision-making, and communication in STEM has been challenging to scale. Real-world simulations require time, coordination, and access to multidisciplinary experts - resources that aren’t always available to every student. AI personas change that equation.
By creating structured, role-based simulations, educators can move beyond passive forms of assessment and toward learning-by-doing environments. Students are no longer simply submitting essays or completing problem sets in isolation. Rather, they’re engaging in dynamic scenarios that require questioning, reasoning, synthesizing information, and justifying decisions. In this sense, Mindjoy’s AI tutors as simulations become both a learning tool and an authentic assessment mechanism.
Importantly, this approach makes experiential learning scalable in ways that were previously impossible. Every student can interact with multiple perspectives, test their reasoning, and refine their thinking in a safe, guided environment. Rather than reserving high-quality simulations for small groups or capstone projects, AI personas allow institutions to embed applied, multidisciplinary practice directly into everyday coursework.
For Mindjoy, this signals a shift from AI as a content-delivery system to AI as a structured simulation engine - one that enables all students to practice complex, real-world problem-solving at scale. As AI becomes more integrated into education, its greatest value may lie not in providing answers, but in creating immersive opportunities to learn by doing.
I'm the founder of Mindjoy 🤓. I'm curious about learning, startups, resilience, systems thinking and the quest for the master algorithm. I wish to create learning contexts that kids call hard fun!