The Future of Higher Education in the Age of AI

Blue book

I’ve been asked by several colleagues for my thoughts on the future of higher education in the age of AI, especially asynchronous courses.

First, as things now are, there is no way to ensure that any assessment completed by students outside the closely-monitored classroom was written by the student. Even before AI, that was the case. For example, I have had students hire others to complete their assignments. AI just makes this kind of cheating available to everyone. Although students who pay for AI may get a higher quality of writing.

Do you believe you can accurately detect AI-written text? There is a new Google Chrome extension that will make AI-generated text sound more human (Iovine, 2026). Sinceerly will get rid of the AI tells, such as commonly-used AI phrases and the em-dash (in almost every conversation I’ve had about this, someone will defend their use of em-dashes; I, too, am an em-dash fan, and I am very human). Since “too err is human,” Sinceerly will add in some typos. (If they’re intentional, are they still typos?) And that’s why Sinceerly is spelled incorrectly. Sinceerly works in your Gmail compose window according to one of three levels you choose: subtle, human, or CEO. Okay, I kind of love that CEO is different from human. And this isn’t the only tool like this out there. The most popular AI tools can be instructed to rewrite their own generated text in a different style, such as something more typical of a college freshman. The user might have to request the typos, though.

Earlier this month, Advait Paliwal created a bot called Einstein (gifted article) (Shroff, 2026). All a student had to do was give Einstein their Canvas username and password, and Einstein would monitor Canvas for new assignments and then complete them. The Atlantic journalist enrolled in a free Intro to Stats class to give Einstein a spin. The journalist aced the course. Or, more accurately, Einstein aced the course. Paliwal insists that he created Einstein to be provocative, to show what is happening right now and what is to come. Paliwal took Einstein down “after he received multiple cease-and-desist letters, including one from Canvas’s parent company” (Shroff, 2026). Paliwal is right. He was just the messenger. “The latest bots have massive context windows, meaning that students can feed in mountains of course content such as syllabi, lecture slides, and practice exams. Today’s agentic tools can complete all kinds of tasks, such as participating in online discussion forums and taking notes on recorded lectures without student intervention” (Shroff, 2026).

What about oral exams? While it’s more labor intensive, some online instructors are meeting with students online using tools like Zoom. When an instructor asks a student an oral question, instructors believe the response is the student’s response. Maybe it is. aiApply’s app, Interview Buddy, is one of several tools available to job interviewees for real-time help. “Our AI tool offers instant answers to interview questions, providing immediate guidance even for unexpected questions” (AI Interview Answer Buddy – Real-Time Interview Assistance, 2026). It’s not difficult to imagine a tool where a student could give AI access to recorded lectures and the textbook (ebook or scanned pdfs), and the AI tool would listen to the instructor’s oral exam questions and generate an answer in real time. In fact, Google’s Notebook LLM might be able to do that right now. Certainly a student could enter all of those data sources into Notebook LLM. The challenge would be running the audio from Zoom (or similar) into the chat box, but that doesn’t seem like an unsurmountable problem. Even if the student knew that they were going to have screen share, they could meet with their instructor using the Zoom app on their phone and have their laptop set up on their desk right behind the phone.  

I’ve spoken with many instructors who have given up on trying to outwit AI and its users. Instead, they say, “The students who want to learn will learn. Those who don’t will use AI.” The big problem with this thinking is that the value of the degree is plummeting. If those who are not interested in learning are using AI to do their work, they will walk away with a grade that does not reflect what they learned in the course. If students use AI in all of their courses, their 4.0 GPA is meaningless, as their new employer will soon learn, assuming this newly minted graduate uses Interview Buddy to fool an employer into hiring them. 

Instructors have used assignments, quizzes, and exams to encourage even the most uninterested students to learn course content and develop important skills. Extrinsic motivation at its best. You complete this work to a satisfactory level, and I’ll give you points. Now, the most uninterested students have found a way to complete the work without actually doing any work, yet they still receive the points.

Scott K. Johnson, an Earth science instructor, “overheard some college students talking about their classes. One was complaining about an assignment they needed to do that night, and another incredulously asked why they wouldn’t just have ChatGPT do it. The first replied, ‘This is my major, I actually need to learn stuff in this class. I use AI for my other classes.’ (Johnson, 2026).

“I actually need to learn stuff in this class”. For the students who want to learn—for those who have intrinsic motivation to learn—they will work to learn. Let’s make our assignments and quizzes optional. Truly optional. No points attached. The students who want to use those assignments and quizzes as tools to learn will complete them. Those who just want the points will not. (There would likely be a few who would do these optional activities just to look good to their instructor—a different kind of extrinsic motivation: impression management.) Instructors would be able to spend more time giving feedback to the students who want to learn—and who would be more likely attend to and use the feedback. [See all of those em-dashes? That’s me, not AI.]

There is one big issue remaining, however. At the end of the term, instructors still need to record grades for all of their enrolled students.

The only solution I can see is a return to well-proctored, in-person, paper-and-pencil exams. This is easy for multiple-choice questions. Instructors of a certain age will need to dust off their knowledge of old-school cheating techniques, such as detecting writing on the back of a soda bottle label. For handwritten essays, blue books started their resurgence in the 2024-2025 academic year (Bereny, 2025). One colleague who has returned to in-class written exams has noticed that the handwriting of today’s students is atrocious. That’s unsurprising given how little practice they get at handwriting. An alternative is a computer classroom where the computers are not connected to the Internet, as one college has done. Essays are typed, printed in the room, and handed to the instructor. The student’s course grade is determined solely by these exams. Students who do the optional assignments and quizzes, though, are expected to do better on the exams because they will have, ostensibly, better learned the material or, perhaps, those students study more because they are intrinsically motivated. It’s an empirical question.

For face-to-face classes, returning to in-person exams is not a big lift. But what about online courses? Higher education accrediting bodies are going to start asking—or should start asking—administrators and faculty what they are doing to protect the integrity of their online degrees. When colleges and universities first introduced distance learning (raise your hand if you remember correspondence courses), students in those courses sometimes had to take in-person exams delivered by an approved proctor. For example, students in the Army could ask their sergeant to be their proctor. In the AI era, students could come to campus and take their exam in a testing center. (Colleges and universities would have to devote funding to expand testing center capacity.) For most students who live locally, coming to campus might not be too much of a burden. For students who have difficulty coming to campus (for reasons such as distance, disability, or caring for someone in the home), proctors (employees of the institution or contractors) could be sent to them. New Mexico State University (NMSU) is in Las Cruces, which is in the southernmost part of the state. Some of NMSU’s students live in the northwest corner of the state, in communities like Shiprock, Gallup, and Farmington, each of which is a 5- or 6-hour drive away. NMSU could partner with local colleges or public libraries to provide proctoring services to their students in those communities.

We cannot put the AI genie back in the bottle. We’ve returned to vinyl LPs (which I get), cassettes (which I don’t get), film in cameras (which I kind of get), and print books (which I also get). It’s time to return to proctored exams while also fostering intrinsic motivation to learn.

References

AI Interview Answer Buddy – real-time interview assistance. (2026). aiApply. https://aiapply.co/interview-answer-buddy

Bereny, A. (2025, November 6). Blue books are back: The revival of pen and paper exams. The Daily Cardinal. https://www.dailycardinal.com/article/2025/11/blue-books-are-back-the-revival-of-pen-and-paper-exams

Iovine, A. (2026, April 25). Sinceerly is an AI tool to “un-AI” your writing. Mashable. https://mashable.com/article/sinceerly-ai-tool-to-undo-ai-writing

Johnson, S. K. (2026, April 13). To teach in the time of ChatGPT is to know pain. Ars Technica. https://arstechnica.com/science/2026/04/to-teach-in-the-time-of-chatgpt-is-to-know-pain/

Shroff, L. (2026, April 10). Is schoolwork optional now? The Atlantic. https://www.theatlantic.com/technology/2026/04/ai-agents-school-education/686754/

image_print