Beyond ChatGPT: How AI Transformed My Classroom Into a Living Laboratory
How 22 student forum posts became a foundation for an experiment in AI's educational potential—and what it reveals about the future of learning
This week, while my Introductory Information Systems students shared eight nouns to describe themselves in an online forum, I ran an unplanned pilot—AI analysis of every word—to map our classroom’s "social DNA." What was a simple community building exercise turned into a demonstration of AI’s potential for pedagogical insight.
The Accidental Experiment
What started as a routine "get to know each other" assignment became an unplanned demonstration of AI's potential. Here's what happened:
The Setup: 22 students posted eight nouns describing themselves, with explanations of personal relevance. Standard stuff—athlete, musician, brother, traveler. The kind of authentic sharing that builds classroom community.
The AI Collaboration: I used Claude Code to enhance some custom tools I had built in my CanvasMCP server so I could extract and analyze discussion forum data. Then I worked with Claude Desktop to connect to the CanvasMCP server to analyze the discussion posts and create interactive visualizations. Two different AI systems, each optimized for different tasks:
Claude Code: Custom CanvasMCP development, data extraction to automate workflows
Claude Desktop: Text analysis, pattern recognition, network visualization, and artifact creation
The Analysis: Together, we identified:
Semantic clusters (who shared similar identity themes)
Communication patterns (whose explanations generated peer engagement)
Community structures (natural collaboration groups)
Individual uniqueness markers (distinctive voices and perspectives)
The Results: The AI revealed 47 meaningful connections between students, extracted 6 distinct personality clusters, and predicted collaboration compatibility—all from analyzing the language patterns in their self-introductions.
What AI "Saw" That I Missed
As an instructor, I read every post. I noticed the athletes, the family-oriented students, the creatives. But AI revealed patterns invisible to me. Of course, this analysis is based on 22 student responses so the cluster labels and relationships are generated/inferred on this small sample size. This cues up a good reflection exercise for students on issues with use of AI.
The Hidden Patterns
The algorithm interpreted that students who used certain noun combinations were more likely to receive peer responses. "Listener" + "Teammate" generated more engagement than "Athlete" + "Competitor." The AI interpreted that vulnerability and connection-oriented language sparked community building better than achievement-focused descriptions, though I would hold back on such interpretations given the sample size. The algorithm identified and labeled students (bridge builders) who served as "connectors" between different personality clusters. Students using abstract nouns ("Energy," "Builder," "Strategist") showed distinct linguistic patterns associated with entrepreneurial thinking. The AI flagged these as potential innovation leaders before I consciously noticed their creative communication styles.
With larger classes, the discussions/documents would multiply, but it is still cheap to analyze. This connects directly to what I explored in my earlier piece on "Intelligence too cheap to meter"—when AI analysis costs pennies instead of dollars, we can afford to run semantic analysis on every student interaction. What used to require expensive sentiment analysis services now costs fractions of cents per student post. We do need to do this keeping in mind the validity of the analysis and sample size.
This wasn't just a cool tech demo. It revealed something fundamental about learning in the AI age: We're teaching students who will work alongside artificial intelligence, so why aren't we learning alongside it too?
Traditional Approach:
Instructor reads forum posts
Makes subjective observations about engagement
Forms intuitive impressions about student personalities
Creates groups based on limited, biased human perception
AI-Enhanced Approach:
AI analyzes semantic patterns and engagement data
Reveals objective connection opportunities
Identifies hidden collaboration potential
Provides data-driven insights for team formation
The difference? The AI saw patterns across 3,000+ words that I couldn't possibly process systematically. It detected subtle linguistic cues that could predict successful collaboration. It identified students who might never have connected organically but share compatibility markers. Though this analysis might be more relevant for larger classes where we might need more effort to create a community.
Potential Applications
This experiment revealed three concrete ways AI can enhance teaching and learning:
1. Invisible Assessment
Instead of surveys or forced personality tests, AI can analyze authentic student work to understand learning styles, collaboration preferences, and engagement patterns. Students learn naturally while AI builds comprehensive profiles for personalized instruction.
2. Data-Driven Community Building
Rather than random group assignments or surface-level similarities, AI could identify optimal collaboration combinations based on communication compatibility, complementary strengths, and shared value patterns detected in student writing.
3. Predictive Intervention
With a bigger sample size, could AI spot students with low peer engagement before they became disengaged. Early identification of connection challenges would allow for proactive support rather than reactive intervention.
The Bigger Picture: Preparing Students for AI-Integrated Careers
My students aren't just learning about information systems—they're experiencing how AI transforms human collaboration. This matters because:
Every Future Workplace Will Use AI for:
Team formation and optimization
Communication pattern analysis
Collaboration enhancement
Performance prediction
Cultural assessment
Students Need to Understand:
How AI analyzes human communication and investigate limitations/biases.
What data patterns reveal about collaboration
How to work with AI insights rather than against them
The difference between AI augmentation and replacement
The Ethical Dimension
This experiment also raised important questions:
Bias: AI patterns reflect training data biases. Could algorithmic team formation perpetuate or challenge existing social hierarchies?
Agency: If AI optimizes collaboration, do students lose the valuable experience of learning to work with diverse, challenging partners?
Transparency: Should students always know when AI is analyzing their work? What about the observer effect changing natural behavior?
These aren't roadblocks—they're essential conversations for educators in the AI age.
The Path Forward: Three Action Steps for Educators
Start Small, Think Big
Begin with simple AI analysis of existing student work. Don't create new assignments—enhance your understanding of current activities. Tools like Claude Desktop or most LLM based chat tools can help with sentiment analysis and pattern recognition, while Claude Code can help build tools to automate data extraction from learning management systems. Start with what you have.
Make It Pedagogical
Don't just use AI for administrative efficiency. Show students how AI analyzes their work, what patterns it detects, and how those insights connect to course concepts. Make the AI analysis part of the learning experience, not hidden infrastructure. I created an interactive "Class DNA Network Map" that students could explore to see their connections—turning AI analysis into experiential learning.
Or flip the script and have students create interactive artifacts as I explored in the article below.
Build it and they will learn
When OpenAI launched ChatGPT with tasks, I setup a task for it to email me a SQL command with explanation and an example every day and keep track of my history. Each day at 11 am I get an email prompting me to learning something about SQL for a few minutes and then click a link to continue to a chat.
What used to require professional developers and weeks of work now takes an educator with a good idea and Claude's artifact creation capabilities.
Address the Ethics Early and Transparently
Build conversations about AI analysis, data privacy, and algorithmic decision-making into your curriculum. Be explicit about when and how you're using AI. Students need to understand both the potential and the pitfalls of AI-enhanced collaboration. Model the transparency you want them to practice in their future careers.
The Meta-Moment
This entire experiment was a collaboration between human pedagogy and AI capabilities. Claude Code helped me build the Canvas integrations that made data extraction possible. Claude Desktop analyzed the patterns and created the interactive visualizations that revealed the community insights. I even used Claude Desktop to help structure and refine parts of this article.
As I shared in "We are all app developers now"—I became an app developer in the process of this experiment, creating interactive learning tools from my students' data in hours rather than hiring developers for weeks. The same process that let me build seven apps in five hours enabled me to transform raw forum data into engaging learning experiences. If you are an educator, you can build and experiment on the fly too.
But here's what's crucial: The AI didn't replace my teaching—it amplified my ability to see and understand my students.
I still designed the assignment. I still made the pedagogical choices. I still interpreted the educational significance. The AI provided capabilities I didn't have: systematic pattern analysis, network visualization, and the ability to process semantic relationships across hundreds of data points simultaneously.
Our students will enter careers where AI is a standard collaboration tool. Shouldn't their education prepare them for that reality by modeling transparent human-AI collaboration?
The Question That Matters
We're at an inflection point in education. AI will either happen to us or with us. The choice is whether we proactively integrate AI as a learning partner or reactively respond to its disruption of traditional pedagogy.
The question isn't whether AI will transform education.
The question is whether educators will transform alongside it.
What's your experience with AI in educational contexts? I'd love to hear how you're thinking about the intersection of artificial intelligence and authentic learning. Hit reply or share your thoughts in the comments.
For more explorations of AI's impact on work and learning, subscribe to stay updated on practical experiments and thoughtful analysis.