Adapting Engineering Teaching for the AI Era: Three Research-Backed Strategies That Work
By zyBooks Engineering Team
Introduction
Students are scoring perfectly on homework, then failing exams when they can’t use AI. They’re submitting polished problem sets without understanding the underlying engineering concepts. When you ask them to explain their approach, they can’t because they didn’t actually work through the problems themselves.
Just as engineering programs eventually accepted calculators, then computers, we now face another shift: students have AI tools that can solve thermodynamics problems, generate circuit designs, and produce complete engineering analyses in seconds.
The core question hasn’t changed: “How do we know what students actually know?” AI just brings urgency to this fundamental assessment challenge.
Three research-backed strategies address this: experiential learning that emphasizes hands-on problem-solving, collaborative work that makes students explain their reasoning, and strategic use of AI to handle administrative tasks. These aren’t new pedagogies. They’re established approaches backed by educational research. What’s changed is their necessity.
Strategy 1: Experiential Learning
Experiential learning, based on Kolb’s foundational framework, shows that hands-on engagement builds deeper understanding than passive content consumption. Design projects requiring physical prototyping force students to grapple with real constraints: material properties, manufacturing limitations, performance requirements. They must understand how theory connects to practice, not just produce correct calculations.
Lab activities where students collect data, operate equipment, and troubleshoot experiments develop skills that can’t be outsourced. Students learn through direct experience with the phenomena they’re studying.
The Shift to Evaluation
Even when students use AI for initial calculations or designs, the learning focus shifts to evaluation at higher levels of Bloom’s taxonomy. Students must:
- Assess whether AI-generated solutions meet specifications
- Evaluate multiple approaches and justify their choice
- Debug problems when theory doesn’t match experimental results
- Explain why design decisions make engineering sense
This shift from execution to evaluation applies across disciplines. Whether in thermodynamics, circuits, dynamics, or materials science, the question changes from “can you solve this?” to “can you evaluate whether this solution is correct and appropriate?”
Strategy 2: Collaborative Learning
Group work creates accountability that individual AI use doesn’t provide. When students must explain reasoning to peers, they catch their own misconceptions. When they hear alternative approaches, they recognize gaps in understanding.
One Team Member’s Evolution
Our approach evolved through experience. Initially: “I wanted students to work together,” designing group projects based on intuition. Then: “When I looked into the research…” everything changed. Educational research on collaborative learning revealed specific frameworks that make group work effective rather than just chaotic.
The difference: moving from “students should work together” to “here’s how to structure collaboration so learning actually happens.”
Practical Implementation
Pre-work creates context. Assign reading or preparatory problems before collaborative activities. Students arrive with baseline familiarity, dramatically improving discussion quality.
Start simple with group formation. Groups of three based on classroom seating work well. Research shows collaborative learning produces benefits even without optimized grouping. As you develop sophistication, use learning analytics to form balanced groups.
Structure prevents free-riding. Assign roles:
- Facilitator: Keeps group on task, ensures everyone contributes
- Recorder: Documents the group’s approach and reasoning
- Presenter: Explains the group’s solution to the class
Design targeted activities. The sweet spot: 10-15 minute activities focused on a specific concept, structured around a worksheet or problem requiring discussion.
Why AI Can’t Replace This
AI can solve the problem. It can’t facilitate the discussion about why one approach works better than another, or help students recognize that their initial understanding was incomplete. This develops communication skills that engineering employers value and makes engineers valuable in professional practice.
Strategy 3: Administrative Efficiency
AI can handle time-consuming administrative tasks, freeing instructors to focus on teaching and meaningful student interaction.
The Rubric Creation Example
Creating detailed rubrics for weekly quizzes can take hours. Our approach: use AI to generate initial rubric structure in approximately 2 minutes. The instructor then reviews and refines based on their teaching preferences.
Time savings: what took hours now takes 20-30 minutes of refinement time. Quality remains high because instructor expertise shapes the final version.
Real impact: This enables weekly or bi-weekly quizzes previously avoided due to time constraints without requiring additional TA support. More frequent, lower-stakes assessment helps students stay engaged.
What This Enables
When administrative tasks take less time, instructors can:
- Provide more frequent feedback without overwhelming workload
- Spend more time on design problems developing engineering judgment
- Have more office hours for personalized student interaction
The instructor still maintains control: reviewing materials, making assessment decisions, determining appropriateness. AI becomes a time-saving tool under instructor oversight, not a replacement for teaching expertise.
Why These Strategies Work Together
These approaches form a coherent system. Experiential learning shifts focus from generating solutions to evaluating them. Students develop judgment to assess AI outputs. Collaborative learning creates accountability through peer interaction. Students must articulate understanding, revealing gaps that solo AI-assisted work hides. Administrative efficiency creates time to implement the first two strategies effectively.
Historical Context
This isn’t engineering education’s first technology disruption. We faced similar concerns with calculators, widespread computer access, and the MOOC era when mass-produced educational content became freely available.
In each case, we eventually shifted from teaching mechanical execution to teaching judgment and selection. When calculators became standard, we stopped teaching logarithm tables and started teaching when to use which calculation approach. AI requires the same shift, just more dramatically.
What Instructors Observe
Faculty implementing these approaches across engineering disciplines report consistent patterns:
Students persist with problems rather than immediately seeking external solutions. Course evaluations specifically cite structured activities as helpful, noting they “helped build confidence even with no prior experience.”
Students develop better ability to connect theory to engineering problem-solving and can explain their reasoning. The performance gap between students with different preparation levels decreases by course end.
One instructor reported that positive student feedback about structured collaborative learning led to adopting these principles for all assignments. Faculty note the focus shifts from policing AI use to designing experiences where AI doesn’t bypass learning objectives.
Getting Started
Implementation doesn’t require curriculum overhaul. Start small, focus on one area, expand based on results.
Choose one high-struggle topic where students consistently struggle or AI use seems most problematic. Examples: lab reports where students can’t troubleshoot issues, design problems with impractical solutions, or homework-exam performance gaps.
Implement one strategy first. For academic integrity concerns, start with experiential learning emphasizing hands-on work. For disengagement, begin with collaborative activities. For time constraints, start with AI-assisted administrative tasks.
Build on what works. The “low-hanging fruit” approach: add role assignments to existing group work, add hands-on components to one assignment, or use AI to draft one rubric. Small, sustainable changes prove more effective than ambitious overhauls.
Expand systematically. After success in one context, apply the same approach to additional topics, then add a second strategy. Build gradually over multiple semesters.
Conclusion
Engineering education has successfully integrated calculators, computers, simulation software, and widely available online content. AI requires the same adaptation, just more urgently.
Students need us to teach engineering judgment—the ability to evaluate solutions, make design decisions under uncertainty, and understand when theory works in practice. They need hands-on skills from working with real systems, equipment, and materials. They need collaborative abilities that make technical expertise valuable in professional practice.
The goal isn’t preventing AI use—that’s neither possible nor desirable. The goal is ensuring students use AI with judgment to evaluate outputs critically, foundational understanding to know when AI helps versus hinders, and practical skills to verify solutions work in the real world.
Start with one strategy in one course. Observe what works. Refine and expand.
Engineering education will adapt to AI by focusing on enduring skills that make engineers valuable regardless of available tools.
The zyBooks Engineering Team develops interactive learning materials for engineering education and works with faculty nationwide to implement research-backed teaching strategies.
