TLDR: A study on the ‘AI User’ curriculum reveals how community college instructors perceive experiential, no-code AI learning activities. The curriculum, designed for non-STEM audiences, uses interactive, scenario-based projects to teach core AI concepts like natural language processing, computer vision, and responsible AI. Instructors highly valued the curriculum’s ability to foster critical thinking, engagement, and real-world understanding through exploration, decision-making, and role-based simulations. However, they also highlighted design trade-offs, emphasizing the need for clear guidance and scaffolding to balance learner autonomy with managing cognitive load in complex tasks.
As artificial intelligence (AI) continues to influence various aspects of our lives, from education to healthcare, there’s a growing recognition that AI literacy shouldn’t be limited to computer science experts. However, many existing methods for teaching AI rely on complex programming or abstract lectures, making them inaccessible to a broader audience, especially those outside of STEM fields.
A recent study by researchers from Carnegie Mellon University introduces and evaluates an innovative approach to AI education called “AI User.” This modular, web-based curriculum aims to teach fundamental AI concepts through interactive, no-code projects that are rooted in real-world situations. The curriculum features eight projects, with this particular study focusing on instructor feedback for Projects 5–8, which delve into practical applications like natural language processing, computer vision, AI-assisted decision support, and responsible AI.
What is the AI User Curriculum?
The AI User curriculum is designed to make AI accessible and engaging for non-STEM learners. Instead of coding, it uses scenario-driven, interactive projects where learners take on professional roles, such as a junior consultant or a housing case reviewer, within simulated environments. They perform tasks that involve decision-making, experimenting with AI systems, and reflecting on ethical considerations. For instance, Project 5 explores large language models in industry, Project 6 assesses computer vision for wildlife monitoring, Project 7 applies AI-assisted decision support in housing services, and Project 8 evaluates responsible AI practices in healthcare.
Gathering Instructor Insights
To understand the effectiveness and applicability of AI User, the researchers conducted a study involving fifteen community college instructors. These instructors participated in structured focus groups, where they completed Projects 5–8 as if they were students. They then provided detailed feedback through individual reflections and group discussions, using a framework called Rose, Bud, Thorn (identifying what worked well, what had potential, and what needed improvement). This qualitative approach allowed the researchers to examine how instructors perceived the design, instructional value, and classroom suitability of these experiential activities.
Key Findings: How Instructors Value Experiential AI Learning
Instructors largely praised the AI User activities for their ability to foster critical thinking, learner engagement, and independent problem-solving. They particularly appreciated features like branching logic, adaptive learning pathways, and realistic decision points that encouraged active participation and concept exploration.
-
Exploration and Engagement: Many instructors noted that the tasks encouraged learners to “poke around” and discover answers rather than follow a rigid path. This was seen as highly engaging, especially in activities like Project 5, where learners interacted with simulated team members to gather information about large language models, or Project 8, which involved a “red-teaming” activity with a medical chatbot, prompting spontaneous collaboration among instructors.
-
Decision-Making and Real-World Relevance: Tasks requiring learners to weigh multiple factors and make informed decisions, such as evaluating housing support in Project 7, were highly valued. Instructors felt these tasks mirrored real-world AI decision-making, with visible consequences reinforcing the learning.
-
Learning from Mistakes: The curriculum’s design, which allowed learners to make mistakes, receive feedback, and revise their understanding, was also a significant positive. In Project 6, comparing computer vision models for animal identification, instructors appreciated that learners could try different approaches without penalty, viewing mistakes as valuable learning opportunities.
Role-Based Simulations for Deeper Understanding
The role-based simulations were deemed highly effective in placing AI concepts within realistic contexts. Instructors observed that these tasks helped learners grasp the collaborative nature of AI work and how different professional roles influence the development, use, and evaluation of AI systems. For example, Project 5 effectively illustrated how AI development involves collaboration between technical and non-technical roles. Furthermore, placing learners in diverse roles like a computer vision intern or a housing case reviewer helped them understand how various stakeholders perceive and utilize AI systems across different domains.
Also Read:
- Assessing AI Readiness: A New Approach for the Modern Workforce
- Advancing Urban Planning with AI: Beyond Prediction to Transparent Reasoning
Design Trade-offs and Areas for Improvement
While instructors were generally supportive, they also identified several design trade-offs. These primarily revolved around balancing exploration with clarity, managing complexity, and providing adequate scaffolding:
-
Balancing Exploration with Clarity: Instructors felt that while exploratory learning was effective, it could become confusing without sufficient guidance. They suggested incorporating onboarding, worked examples, or guided walkthroughs to help learners orient themselves before independent exploration, especially in complex tasks like Project 7, where subtle distinctions in housing support decisions could be challenging to interpret.
-
Managing Complexity: Tasks involving multiple similar scenarios with subtle differences, such as in Project 7 and 8, were seen as useful for reinforcing learning but could also be overwhelming if distinctions weren’t clearly framed. Instructors recommended careful curation of case-based complexity, ensuring each variation served a distinct instructional purpose.
-
Supporting Autonomy with Scaffolding: While autonomy was valued, instructors noted that too much could lead to frustration. They recommended clearer instructional support, progressively disclosed feedback, and “bottom-out” hints to guide learners who struggled, ensuring that scaffolding enabled rather than limited learner autonomy.
This study underscores the importance of instructor perspectives in designing effective AI literacy tools. The findings reinforce the value of experiential and inquiry-based learning, particularly for non-STEM audiences, while also providing actionable insights for refining interactive AI learning activities. The AI User curriculum, with its focus on real-world scenarios and no-code interaction, represents a significant step towards making AI education more inclusive and impactful. You can learn more about this research by reading the full paper here.
