AI, Collaboration, and Changing Classrooms: Fresh Ideas on Teaching Design from WKU’s Workshop When...
The "Black Box" in the Classroom: Understanding Student Misconceptions of AI
As generative AI becomes a standard fixture in modern education, a gap is emerging between students’ ability to use these tools and their understanding of how they actually work. While much attention has been given to the benefits of AI-powered learning, less focus has been placed on the assumptions students bring to their use of these tools.
A recent study, “Exploring Students’ Misconceptions about ChatGPT-generated Text” suggests that beneath the surface of effortless essay generation lies a web of technical misunderstandings that could lead to "informed mistrust"—where students trust AI output not out of laziness, but based on a fundamental misinterpretation of how the technology processes information. This newsletter will break down the key findings, particularly focusing on the misconceptions and their implications for teaching and learning.
Key Findings and Implications
The study surveyed 267 second-year high school students from two schools in southeastern Norway. Using open-ended questions and thematic analysis, it found that students tend to adopt a practical, task-oriented approach when using ChatGPT.
At the same time, four major types of misconceptions were identified: (1) information sources, (2) output content, (3) operational mechanisms, and (4) usage behavior.
Table 1. Overview of students’ misconceptions about ChatGPT, based on qualitative analysis of responses
1. Understanding of Information Sources
Although more than half of the students (n=183) recognized that ChatGPT generates text based on internet data, many showed confusion about where this information comes from. A total of 115 students did not fully understand the diversity of sources or the model’s inability to verify accuracy.
What this means for teaching: Students may overestimate the reliability of AI-generated content. This highlights the need to strengthen source evaluation and verification skills when using AI.
Although more than half of the students (n=183) recognized that ChatGPT generates text based on internet data, many showed confusion about where this information comes from. A total of 115 students did not fully understand the diversity of sources or the model’s inability to verify accuracy.
What this means for teaching: Students may overestimate the reliability of AI-generated content. This highlights the need to strengthen source evaluation and verification skills when using AI.
2. Understanding of AI Outputs
Students valued the alignment between AI output and the task most highly (n=81), and some demonstrated awareness of potential inaccuracies (n=62). However, 47 students expressed uncritical trust, treating AI output as authoritative. Others viewed AI primarily as a tool for rewriting (n=45) or summarizing information (n=34).
What this means for teaching: Students are already developing partial critical awareness, but this needs to be extended. AI should be framed not just as an answer generator, but as a tool for supporting thinking, drafting, and revision.
Students valued the alignment between AI output and the task most highly (n=81), and some demonstrated awareness of potential inaccuracies (n=62). However, 47 students expressed uncritical trust, treating AI output as authoritative. Others viewed AI primarily as a tool for rewriting (n=45) or summarizing information (n=34).
What this means for teaching: Students are already developing partial critical awareness, but this needs to be extended. AI should be framed not just as an answer generator, but as a tool for supporting thinking, drafting, and revision.
3. Understanding of How AI Works
A majority of students (n=64) had only a vague association with the term “algorithm,” while only a minority (n=33) understood that the core mechanism is a probability‑based text‑prediction process. Anthropomorphizing AI—treating it as if it “thinks” or “knows”—was a common misconception.
What this means for teaching:AI literacy should go beyond surface-level explanations. Students benefit from a basic conceptual understanding of how large language models generate text, helping them avoid attributing human-like reasoning to AI.
4. Patterns of AI Use
Only a few students (n=7) voluntarily mentioned the need to verify information sources, revealing a clear tension between efficiency orientation and critical verification. Many prioritized efficiency over accuracy, indicating a gap between task completion and critical engagement.
What this means for teaching:Educational practice should emphasize that AI can only serve as an aid for idea generation and text refinement, not as a substitute for independent thinking; students must be trained to habitually verify AI outputs’ sources and factual accuracy.
Only a few students (n=7) voluntarily mentioned the need to verify information sources, revealing a clear tension between efficiency orientation and critical verification. Many prioritized efficiency over accuracy, indicating a gap between task completion and critical engagement.
What this means for teaching:Educational practice should emphasize that AI can only serve as an aid for idea generation and text refinement, not as a substitute for independent thinking; students must be trained to habitually verify AI outputs’ sources and factual accuracy.
Conclusion
Overall, the study shows that students’ misunderstandings span four key areas—information sources, outputs, mechanisms, and patterns of use—and these directly affect their ability to use AI critically. Based on these findings, several directions for teaching emerge:
- Strengthen information literacy by helping students evaluate sources, question outputs, and recognize the limits of AI-generated content
- Guide more intentional use of AI, positioning it as a tool for supporting thinking rather than completing tasks
- Build a basic understanding of how AI works, helping students move beyond anthropomorphic assumptions
- Embed verification and reflection into assignments, making critical evaluation a routine part of learning
Reference:Bråten, I., Latini, N., & Strømsø, H. I. (2026). Exploring students’ (mis)conceptions about ChatGPT-generated text: A qualitative study. Education and Information Technologies. https://doi.org/10.1007/s10639-026-13938-w
Faculty Spotlights
WKU Faculty Share Practices on AI in Teaching and Assessment at UNNC Forum
Three faculty members from Wenzhou-Kean University were invited to present at the Technology-enhanced Teaching Forum 2026, held on 10 April at the University of Nottingham Ningbo China (UNNC).The forum, themed The Assessment Revolution: From Pilots to Practice in the AI Era, brought together educators exploring how generative AI is reshaping teaching and assessment. Representing Wenzhou-Kean University, our faculty shared their experiences and insights on integrating AI into teaching practice, contributing to discussions on moving from experimentation toward sustainable implementation.
Presenter: Shaoshuang Wen
Dr. Shaoshuang Wen from CLA presented “Who Trusts the Machine, Who Resists It? Q-Method Evidence on Student Viewpoints Toward GenAI Writing Feedback in a Crossover Classroom Study.”This study uses Q-methodology to map undergraduate students’ structured viewpoints toward GenAI writing feedback in a counterbalanced crossover classroom design, identifies shared viewpoint types, and offers design implications for feedback practices that foster epistemic agency and responsible GenAI use.
Presenter: Chang Liu
Chang Liu delivered From Syntax to Synergy: Personalized Pathways and AI-Assisted Assessment in Concept-Driven Creative Coding. This presentation examines the AI-driven paradigm shift in visual communication education, guides students to engage AI as a collaborative "logic partner", and details how to restructure pedagogy and evaluation to reflect this philosophical shif, ensuring human creativity remains the ultimate arbiter of value in the AI era.
Presenter: Jie Zhao
Jie Zhao, a student of Svetlana Vikhnevich, presented her work on her behalf due to a scheduling conflict. Drawing from the study Buddy AI: Enhancing Freshman ESL Reading and Vocabulary through Interactive Chatbots, this initiative explores an AI-powered chatbot as a personalized “study buddy” to support freshman English as a Second Language (ESL), enhances vocabulary retention and reading comprehension, and provides practical guidance on designing custom AI chatbot personas for classroom use.
The participation of our faculty in the forum provided an opportunity to engage with current conversations around AI in teaching and assessment.
As AI continues to evolve, these exchanges highlight the importance of exploring how we can design more effective pedagogical approaches and assessment practices. We also welcome colleagues to continue sharing their experiences and perspectives, whether through formal events or informal conversations, as we collectively navigate the opportunities and challenges of AI in education.
CTL Updates
Professional Development Day (PDD)
We are pleased to announce that the Professional Development Day (PDD) is scheduled for Thursday, June 11, 2026. In this regard, you are cordially invited to submit your presentation proposals for consideration.
A wide range of topics is welcome. This year, we especially encourage submissions that explore AI expectations, AI-powered teaching, and AI-enhanced research. If you are interested in presenting at the PDD, please submit your proposedtopic(s) along with an abstract, narrative, estimated duration, or a brief description of your presentation. We will be accepting proposals until Tuesday, April 21, 2026. Please note, proposals received after this deadline will regrettably not be considered. Furthermore, for those who are currently outside China and plan to deliver their presentation virtually, please inform us of the time difference between China and your location to facilitate appropriate scheduling.
Note: The standard presentation time is 30 minutes. This year, we also welcome workshops and seminar-style sessions. If proposed, these may be allocated up to two time slots, totaling one hour.
The selected topics for presentations will be announced to the WKU faculty and staff shortly after the April 21 deadline.
Your willingness to participate in the PDD is highly appreciated. We eagerly anticipate your valuable contributions and look forward to your submissions.
For more information and to view previous PDD events, please visit: https://www.wku.edu.cn/en/ctl/pdd
Registration Now Open | Professional Development Program on Artificial Intelligence Literacy in the Digital-Intelligence Era
The program centers on the integration of AI-driven creativity and pedagogy, offering innovative teaching methodologies and insights into cutting-edge technological developments. The curriculum covers AI-empowered teaching practices, case studies in instructional innovation, and emerging frontiers in AI technologies. Instruction is delivered by senior faculty members of Peking University, supported by doctoral-level teaching and research teams.
-
Dates: April 23–26, 2026
-
Location: Yanyuan Campus, Peking University (Beijing)
-
Format: In-person instruction + visits and study tours + seminars and discussions
-
Language: Chinese
For more information, please click here.
Instructional Skills Workshop
We are pleased to announce the upcoming Instructional Skills Workshop (ISW) — a signature faculty development program designed to strengthen teaching practices through experiential learning.
🗓️ Upcoming Sessions
- Session #26: May 16-17 & May 30-31.
Participants will explore core teaching and learning concepts, reflect on current practices, experiment with new instructional strategies, and engage in supportive environments.
We look forward to discussing this excellent opportunity to increase student engagement with interested faculty and staff of WKU. Please note that participation is limited to five individuals, and all four days of the workshop are required.
Scan the OR code to register >>
Author: Yan (Tanya) Tang
Chief Editor: Yirui (Sandy) Jiang