To AI or Not to AI
Lessons from allowing students to use AI as much as they want
Going back to the classroom
Before the pandemic, Cornell Tech, one of the graduate schools of Cornell University focused on business, technology, and entrepreneurship where I teach, had a policy that discouraged its students from using laptops and smartphones for notetaking in classrooms. Studies showed that taking notes on devices wasn’t as effective in learning as doing so with pen and paper. I appreciated this policy because I could explicitly tell students to put away their devices and not feel bad about it.
During COVID, the university abandoned this policy for obvious reasons and left it up to the professors whether to enforce it when we returned to the physical classroom.
At the beginning of my course in 2024, I announced to my students that they needed to turn off their phones and close their laptops. No one said anything at first, but I could feel the tension in the room rise. Many of the students were born in this millennium. They don’t have much or any memory of the world without laptops, let alone smartphones.
“We are Cornell Tech. We are supposed to use technology.” After a brief moment, one student broke the silence, voicing what everyone else had in mind.
Many others seemed equally discontent with my announcement. I proceeded with the no-device policy anyway. By the end of my first lecture, out of the 50 students in my class, there were more than a dozen students with their laptops and smartphones open, some sheepishly, some shamelessly. Whether they were taking notes, browsing, or checking TikTok, I wouldn’t know. They were mostly in the back of the classroom.
After the first lecture, I decided to do away with this policy.
Instead of manufacturing attention from my students by forcing them to pay attention, I took it upon myself to see if I could make my lecture engaging and interesting enough that they would want to do so. If they weren’t, well, that wasn’t my loss, I reasoned.
In 2024, however, I had bigger fish to fry.
To AI or not to AI?
18 months ago, I didn’t need to worry about my students using AI to do the assignments I gave. It was a different story this year.
For this year’s course, I contemplated banning the use of AI. However, I quickly figured that trying to force students not to use AI was a futile exercise. It’s human nature to find an easier and more convenient way to do things, especially when it comes to things like school assignments.
“Use as much AI as you’d like,” I told my students at the beginning of this semester. “Just be transparent about how you used it in your assignment submission and document it." Even if I told them not to use AI, there would be no sure way of deciphering where and how the students leveraged AI.
Assignments for my course this spring semester, Branding for Products, contained reading, analysis, writing, naming, and some amount of designing. The course is for MBA and Engineering students and the majority of students don’t have any design or creative training. I made the design aspect of the course lighter than I would have liked it to be but enough so that they’d have to use design tools such as Figma.
Key takeaways
1. AI raises the average… but it’s still average.
I expected AI to not only level the playing field for students but also to elevate the performance of all students, including not-so-good ones.
To an extent, it does. For instance, reading and writing.
Before AI, I used to see a lot more range in my students’ work for assignments that included written analysis. Now, while a lot of their work seems much cleaner, clearer, and even better organized, they all sound plastic, mechanical, and impersonal.
I have 50 students to grade (and I am in the midst of grading their Final Projects). By the third or fourth submission, the points, the patterns, the structures, and even styles in their work become very repetitive. Not only do they sound similar, they sound the same. In other words, very average.
Some students were forthcoming about where and how they used ChatGPT or Claude, but many weren’t. Some were even open about feeding the reading material to such tools, getting the summary, and submitting that output, which defeated the very purpose of the assignment itself.
After the first assignment, I realized that the way I phrased each task in the assignments mattered way more than it used to. I could no longer ask them to just read and analyze. I now needed to ask them to think about their own unique personal experiences and incorporate them into the analysis, for instance.
As teachers and professors—and even as parents—the role we play as educators is more about critical thinking than ever. Everyone can now produce similar things at similar quality. What then matters is how your thinking is different from others.
2. Discipline trumps technology
This year, each session of my course was over three hours long (brutal, I know - not my choice) so I had ample time in the classroom to create live dialogue and interactions. In addition, I would hold online discussions in the form of threads in Slack so that those students who might not feel comfortable speaking up in class, i.e. international students whose first language isn’t English, could also participate without intimidation.
I’ve implemented this multimodal participation method for my course and it’s become clear to me that the more quickly they engage in these online discussions, the better they end up doing overall in the course. My course is an elective, meaning that the students choose to, and are not required to, take my course. Not only has this multimodal participation method become a good way to gauge the seriousness of their interest in the subject matter I teach, but it is also a good tell for their discipline.
Before 2024, my students had to respond to these online discussions by doing research, interpreting, and writing themselves. In 2024, I could tell that this process was replaced by copying and pasting the question as a prompt, and then, copying and pasting the answer given by AI.
After the first discussion, I tweaked the way I posted the topics and tasks at hand so that it was tricky for them to use AI. For instance, I would ask them to cite a specific use case scenario and take a screenshot of it to include in their answers. AI isn’t yet capable of understanding that level of nuance so students had to do their own thinking and manual work.
In addition to critical thinking as the core part of teaching, teaching soft skills such as discipline, decision-making, and professionalism is more important than it used to be until the very recent past.
3. Quality is a function of time and care
The work of the students who are letting AI do the work all sounds the same. It lacks original thought. The work that’s done by a real human is less cold, less machine-like. It has a personal touch. Even if it might have mistakes in it, it is more interesting.
I do see evidence of AI making the fidelity of the output better. But that’s entirely different from the quality of their thinking. As of spring 2024, I’m yet to see evidence of AI making the thinking of students better. It’s almost the opposite: it’s making them lazy.
The course I teach is not math. The output isn’t graded based on whether their answers are correct or not. It is based on how insightful or interesting their thinking and the subsequent output are. When AI is used to replace the thinking part of the work, the end output is not only average but bad even.
In the final stretches of grading this semester’s work, the work that has obvious signs of AI tends to be shallow and lazy, and it shows a lack of care. It is quickly obvious that those students didn’t spend the time to think, craft, and finalize the work. The work doesn’t show there is care. What is bubbling to the top shows little signs of AI.
In the age of AI, what is old might be new again, as David Lee, the Chief Creative Officer, and my good friend, put in my conversation with him a little while ago.
Quality is a function of time and care. That is an adage that deserves a renewed reminder for all of us.
4. Taste takes time to learn
From the age of about ten years old, I wanted to be a professional football player.
With my brothers, I would practice in our front yard even after the sun set and play until it was way too dark to see the ball. When I was about twelve, I asked my parents to send me to Brazil so I could play among some of the best in the world. To my parents’ credit, they didn’t say no. Instead, they suggested that I enroll in a boarding school in Shizuoka, a region in Japan that had—and still has—several top high schools that had national-level football teams. Before playing among foreign players, they reasoned, I should play among Japan’s best teenage players and become one myself.
I don’t remember exactly when but I realized I wasn’t born with the talent to be the next Pele, Platini, or Maradona. I just didn’t have it in me.
Unlike athletic talent, however, creative taste is not something we are born with. It’s something we develop over time.
Where athletic talent and creative taste have similarities is how we develop them.
It is a famous tale that top athletes like Michael Jordan and Kobe Bryant had unparalleled discipline and practiced more than anyone on the team. Like athletic talent, creative taste requires a tremendous amount of training, practice, dedication, and time to learn and develop. It is learned over time through exposure, influence, and experience. Taste takes time to develop.
To AI or not to AI? This soon will be an irrelevant question.
In less than ten years, AI will be as ubiquitous as the Internet and smartphones. The business world went from the digital-first mentality to the mobile-first one. Many corporations are now adopting the AI-first mentality. Just like we can’t go about life without the Internet and smartphones, we won’t be able to avoid AI.
We should also know that if we use AI, chances are everyone else is already using it for whatever we are trying to do. That makes us like everyone else. See point No. 1 above.
With AI, mediocrity is now available for free.
If you want to differentiate, before putting AI to work for you, focus on your critical thinking, discipline, care, and taste first.
Without those, you might as well let the machines take over what you do.
Please hit reply with any feedback, thoughts, or questions. I’d love to hear from you.