Two Electrical Engineering Professors Split on AI Assignments

AI assignment – A graduate student interviewing two electrical engineering professors finds starkly different approaches to AI in course and assignment design—one cautious and exam-focused, the other openly integrating AI into projects. Both agree on a central goal: assignmen

The first time one professor said he had proof of AI showing up during assessment, it happened in real time: during a final exam, he saw a student using ChatGPT. He checked the responses against ChatGPT’s output and found they matched.

A year earlier, artificial intelligence had already been on his radar. But that exam was the moment it stopped being an abstract concern and became an academic problem with consequences.

At the same time. in the same broad field of electrical and computer engineering. another professor took a different path—building AI into learning rather than treating it as something to block. In their courses. students use tools like Verilog simulations. write reports tied to ABET outcomes. and now. increasingly. use AI for specific parts of how they think and work.

Their conversation, drawn from a graduate student’s interviews at Baylor University, lays bare a split that many educators are quietly wrestling with as AI tools like ChatGPT and Copilot become widely accessible and flexible.

One professor—referred to as Professor A—teaches three junior and senior level electrical and computer engineering courses: Microprocessor Systems. Computer Organization. and Embedded Systems Design. The courses lean heavily on programming. In Microprocessor Systems and Embedded Systems. students submit assignments that require code. and then the code is run to confirm it works. In Computer Organization. labs ask students to write code to design a processor. paired with more traditional Canvas homework: mathematic-type problems and logic problems.

Professor B also focuses on electrical and computer engineering, lately emphasizing digital logic. They have taught an Intro to AI course as well. In that work, students do hands-on projects in Verilog, run simulations, and write short reports connected to ABET outcomes.

When both professors were asked when they first noticed students using AI tools. Professor A described the shift from general awareness to certainty. He said he became aware “a couple of years ago” but it became clear about a year ago. in Fall 2024. when he saw a student using ChatGPT during a final exam and verified that the answers matched.

Professor B described a faster adoption of that awareness, saying they were an early adopter. They noticed student use right away, and they recognized it in the way some writing and code explanations suddenly looked “too polished or generic.”

Their attitudes toward AI in education captured the central divergence.

Professor A said his view is “both” opportunities and challenges. He is not anti-AI, but cautious. In his account, students need to learn core principles first and to solve problems themselves before relying on tools. He compared it to giving a first grader a calculator: the answers might be correct. but the student doesn’t understand the math behind them.

Still, he added a practical example of where he thinks AI can help once fundamentals are in place. In his Embedded Systems class. during the final project. he actually encourages students to use AI to help with features not formally covered. like adding a touchscreen display. In that setting, he said AI accelerates creativity and implementation.

Professor B’s overall stance was “mostly positive.” They said AI is a major opportunity if used intentionally, but it needs structure so students still do the thinking.

That difference shows up most sharply in how the professors respond to assessment.

When asked whether they modified assignments to make courses more “AI-resistant” or “AI-integrated,” Professor A described limits. For homework, he said there isn’t much he can do to fully prevent AI use, so he has to trust students to follow guidelines.

But for exams, he changed the process. He now uses Respondus Lockdown Browser. He said he doesn’t particularly like it because it can be glitchy. but the reason is clear: he said he had students use ChatGPT during exams “two semesters in a row. ” and both of those students failed the course for academic dishonesty. His goal is simple in his telling—remove the temptation.

Professor B described the opposite approach: instead of trying to block AI, they build it in. In their example, students might use ChatGPT to explain or refactor their code, then critique what it got wrong.

Both professors were also asked to compare the current moment with the earlier wave of assignment changes during COVID, especially in online settings.

Professor A said that during COVID. many students clearly didn’t learn as much for various reasons. and that it took time for academic performance to recover. His takeaway now is that the same basic risk persists: if AI is used in place of learning—not after learning—students won’t develop the competence needed to function as engineers. For him, the purpose is preparation for real jobs, not just helping students pass classes.

Professor B said the worry during COVID had been engagement. Now, they said, the worry is authenticity. They described both periods as pressure to design assignments that show students’ thought process, not just the final product.

Their accounts also touch on how institutional guidance has shaped their decisions.

Professor A said the Provost’s office provides guidance and suggested syllabus language. and that the topic was discussed at a school-wide faculty retreat. At the department level, he said he doesn’t recall formal discussions. He added that campus-wide guidance has been sufficient, though he still wants more tools and clarity—especially for code evaluation.

Professor B said Baylor encourages faculty to set their own course policies. In their view, that means acknowledging AI and using it responsibly.

Collaboration with colleagues is where the tension inside the same department becomes visible.

Professor A said he has discussed AI approaches with one colleague, and that they do not agree. His colleague believes AI use is inevitable and that students should be taught to use it from the beginning. Professor A said he believes students should first learn to solve problems without it and only then incorporate AI later. “We agree to disagree,” he said.

Professor B described collaboration of a different kind. They said they have worked with colleagues to rethink assignments and even automate parts of grading with AI while keeping instructor oversight.

Both professors closed with a theme that sounds less like a compromise than a shared necessity.

Professor A framed it as balance: students need foundational skills to make AI useful rather than limiting. He also said he wants to stay open to change. If the academic world evolves and the best way to prepare students changes. he says he should adapt—yet for now. he believes students must learn core principles first.

Professor B echoed a parallel certainty in a different direction: AI isn’t going away, and that’s fine. The goal now, they said, is to help students learn “with it,” not around it.

What comes through in the interviews is not a single right answer, but a question that has become impossible to ignore for instructors across disciplines: how do you design assignments when AI is constantly present?

Professor A’s focus is on preventing AI from substituting for learning during the moment students are supposed to demonstrate knowledge. Even though he encourages AI in a structured way during projects—like helping with touchscreen features he hasn’t formally covered—he draws a line at tests and uses Respondus Lockdown Browser after two semesters of ChatGPT-related academic dishonesty.

Professor B’s focus is on turning AI from a threat into a learning instrument. They described using ChatGPT to explain or refactor code. then requiring critique of what the tool got wrong—so the student’s thinking stays visible. They also described a broader effort to incorporate AI intentionally through course structures and even automated parts of grading. while maintaining instructor oversight.

Between those positions lies the students’ real experience: some will learn under rules designed to restrict AI at the point of assessment. while others will learn under assignments that ask them to use AI and then interrogate it. Either way. their work is now being shaped by the same new reality—ChatGPT. Copilot. and other AI tools that can generate flexible answers and. increasingly. tempt students to skip the hard part.

The graduate student who conducted the interviews, Luke Mello, described both professors as teaching from their own experience and beliefs, illustrating how even in the same discipline educators can design courses and assignments differently when AI is part of the classroom.

Behind the personal contrast. the broader literature referenced alongside the interviews paints an academic backdrop: Uysalel’s work on using AI interactive interfaces to enhance student engagement; Barakat’s argument for intentional AI incorporation in the engineering curriculum with critical evaluation of output; and Jiménez Romanillos and Andersson’s exploration of Bloom’s Taxonomy in design education. showing how AI can reshape the cognitive expectations of assignments.

For now, the two professors’ disagreement remains practical and grounded in daily teaching decisions—what students submit, how they’re tested, and what instructors choose to trust.

In a world where AI is accessible enough to be used during exams and sophisticated enough to help with new features in a final project. the stakes are no longer hypothetical. They show up in a proctored moment. a failed grade. a student’s reported “polished” work. and the quiet work of redesigning assignments so learning stays at the center of engineering education.

AI in education ChatGPT Respondus Lockdown Browser engineering curriculum assignment design Baylor University electrical and computer engineering academic integrity ABET outcomes digital logic Verilog