How Generative AI Owns Higher Education. Now What?
Let’s assume that you’re “teaching” a course. Now let’s assume that you’re taking a course or have enrolled in a degree program. Let’s also assume that you know something about generative AI (GenAI) which many professors and lot’s more students already do. Let’s also assume that GenAI tools will become incredibly smarter, better and faster, which is the easiest prediction anyone can make.
If all of the above assumptions are true – and they obviously are – what happens to higher education?
Have we missed something huge here?
We sure have.
If fact, it’s so big that it literally changes the very premise of learning as we know it today. It’s astonishing that most professors, administrators and even donors don’t see the proverbial train barreling down the tracks, perhaps like how climate deniers cannot calculate storms, rainfall, floods, droughts and unbearable summers. Maybe it’s just a repeat of the Luddite phenomenon that surrounds the adoption of all new technology. Who knows, but worse, many universities have actually banned GenAI which is a naïve attempt to regulate a technology more compelling than the Internet and in so doing have actually provided encouragement to faculty and administrators to pretend that GenAI and its CustomGPT children are more of a threat than a service. Further, faculty who have not used GenAI to develop and deliver courses have already missed an extraordinary opportunity to improve the learning process for their students and increase the learning outcomes of their courses.
As a professor of business technology, I have begun to treat GenAI and CustomGPTs as willing teaching assistants only to discover that they’re much closer to partners than assistants. I have asked Gemini and ChatGPT (and others) to develop syllabi and then compared them with my own as a way to improve my courses. Gemini and Chat have often made terrific suggestions and found materials I missed in what I thought was an exhaustive search. Selected CustomGPTs also summarize huge amounts of text, website content and even videos which also helps enormously. Next step: use CustomGPTs to create course videos from textual or spoken requirements which will be especially helpful for fully online courses — in multiple languages.
Course Design & Development
Let’s start with a course – a graduate course for our purposes here – in marketing. Many professors in the not too distant past – and definitely even today – develop their syllabi by converting research and knowledge (and sometimes even actual experience) into what they believe represent marketing principles, cases and best practices with a wide variety of readings, content, videos, etc. Textbooks are still used as the field itself – all fields – evolve faster than publication schedules and where many academics still frown upon applied versus theoretical research. If the truth be told, professors cannot possibly track all of the basic and applied research in marketing (or any field) published across the globe, in spite of what Google Scholar says. Few have the time to read all of this research. That said, and of course depending on the professor, traditional, “unassisted” courses can generate some useful learning outcomes.
But what about now?
Let’s look at how large language models can develop a syllabus with detailed prompts like this:
“Develop a syllabus for a marketing course for graduate students that covers the fundamental principles of marketing, marketing cases, readings that include some theory and practice – with an emphasis on practice – with requirements that include projects, essays, tests and in-class conversations that illuminate theory and practice. Also develop lecture notes for me with bullet points – derived from the readings – that will focus the students class-by-class on the right topics. Please also assume that the class is 15 weeks long with readings, assignments and 5 salient topics (captured in the bullet points) per week. Use only current readings: nothing older that 2015. I’d also like some learning outcomes I should expect the course to generate.”
If you doubt the power of GenAI, prompt Gemini with the above and then follow-up with: “please provide a longer list of readings including articles and books as well as links to web sites that support the course.”
Play with Gemini a little more by asking: “please identify some online company and YouTube videos that can be used each week.”
By the way, I did all this in less than 3 minutes. Anyone can. (Imagine what could be developed in 30 minutes.)
What about course videos? Professors can create them (by lecturing into a camera for several hours hopefully in different clothes) from the readings, from their interpretations of the readings, from their own case experiences – from anything they like. But now professors can direct the creation of the videos by talking – actually describing – to a CustomGPT about what they’d like the video to communicate with their or another image. Wait. What? They can make a video by talking to a CustomGPT and even select the image they want the “actor” to use? Yes. They can also add a British accent and insert some (GenAI-developed) jokes into the videos if they like. All this and much more is now possible. This means that a professor can specify how long the video should be, what sources should be consulted and describe the demeanor the professor wants the video to project.
This only touches the surface of what “professors” can do with GenAI and CustomGPTs to create courses with everything – and a whole lot more – they need or want to teach a great marketing course, which fundamentally changes the way courses are conceived, designed, developed and delivered in higher education today. Did I forget the cases? Introduce a product and then ask Gemini to write the press release and describe the marketing campaign. If that’s a student assignment, it only takes 2 minutes to complete the task.
It’s gets worse – or better, depending on one’s perspective. Many professors assign readings and then discuss them in class. But what if the readings were summarized and then interpreted by GenAI. What if – and hopefully this isn’t the case today – a professor wanted a PowerPoint presentation generated from the readings (not textbook readings – they already come with dated PowerPoints). No, I’m talking about random readings – maybe even something professors wanted the students to immediately read or watch – like something about Superbowl ads or the endless stream of Ozempic commercials. Yes, regardless of its source or format, it could be summarized, interpreted and packaged for presentation without any professorial touch.
What’s a “Professor”?
With all this power, what’s the unique contribution professors can really make? Arguments are made all the time about how effective classroom experiences can be, that there’s something special about human-to-human contact in the learning process.
Everyone likes to reassure professors that they’ll always be necessary, but will they? It may be that accreditation boards will save them from what otherwise is inevitable – which is replacement along some assist/partner/replace continuum over a period of time, but no later than 2035. In 2024, Gemini has already reduced their contributions to a 3-minute waltz.
What happens when syllabi are easily better than anything professors could design or develop? When, for example, will GenAI read MRIs faster and better than radiologists? Conduct more accurate breast examinations? Admit/reject students to colleges and universities? There are trends here that cannot be ignored. It’s all just around the corner. Professors are not a protected species.
What about when more and more courses move online? Professors who teach online are barely there anyway, right? (You’d think that it would occur to university professors who host online courses today that they’re demonstrating how unnecessary they are to the learning process. They develop syllabi, identify content, develop [or appropriate someone else’s] videos, identify assignments and tests – all of which we’ve already demonstrated can today be enabled by GenAI. Fully asynchronous students often have little or no human contact with actual professors. How does all that demonstrate necessity?)
GenAI can develop curricula without any human touch and deliver courses – content, videos, exams, etc. – also without any human touch. In fact, while accreditation boards would never approve, it’s absolutely possible to automate the entire online education process with GenAI and CustomGPTs. Professors nod approval (for the sake of accreditation) of the courses they do not create and the grades they no longer give. Crazy? Not at all. In fact, it’s possible right now.
Students, GenAI & Custom GPTs
Let’s take the content and the requirements of the course – the video lectures, readings, research reports, websites and essay examinations – and match these activities with GenAI and Custom GPTs. First, students do not need to read or watch anything. Let’s say that again: students do not need to read or watch anything – unless their professor takes roll – even if their professor takes roll. Instead, they can rely upon summarizers and converters to reduce their workload. If they choose to avoid summarizers and converters, they can hire GenAI tutors to see how well they’re doing and suggest how they might improve their performance. Regardless, they no longer have to read articles, inspect web sites, immerse themselves in textbooks, write essays or take tests.
With the help of GenAI/CustomGPTs, online students actually have very little to do. Summaries of everything are easily generated, and student requirements are almost as easily satisfied. They too don’t need to read or watch anything. In completely asynchronous courses, no one even checks, even if videos can determine if they’ve been watched. In fact, students may appear to have read and watched everything – as evidenced in the tests “they” take – when they’ve barely bonded with course materials.
What’s a “Student”?
When students have all this help, what’s their role in the learning process? First, let’s assume they will use all the help they can find because it’s unlikely students will reject ways to learn faster and easier. Part-time students will especially appreciate short-cuts, since they’re on a different educational clock than full-time students — who will also accept the help. But it’s not at all clear how learning outcomes will be measured, unless learning outcome metrics are exchanged for others, like speed/ease-to-degree-completion. For students, the assignment is to understand the relationship between real versus surrogacy, and how “engagement” should be defined — that ends with an “A” for the course.
Now What?
Academia’s response to all this has been uneven. One the one hand, universities are warning – and punishing – students who use GenAI tools to satisfy course requirements. But on the other hand, they’re teaching the technology where students are required to use GenAI to satisfy course requirements. It’s pretty quiet about how it expects faculty to deal with GenAI and especially CustomGPTs.
There’s no question that academia’s response is late, inconsistent and incomplete. The world of CustomGPTs has acerbated the challenge, as more and more special purpose education applications help faculty and students reduce their contributions to the learning process.
A wide open question: “who does what?” Another one: “why are professors compensated the way they are when their digital assistants and partners do much of the work?” And, lastly: “how can anyone measure learning outcomes when they’re contrived by GenAI and CustomGPTs?”
Higher education needs a fast audit. The roles that professors and students should play in the education process must be re-defined – and then re-invented, if not re-imagined altogether. The larger issue is the relationship between humans and AI. It’s essential that professors and students understand how to work with increasingly intelligent machines that will evolve from “teaching assistants,” to “faculty partners” and eventually to “curriculum bosses,” and how faculty will share – and eventually yield – decision-making and problem-solving power over time. As suggested, students will define their roles as team members where courses are taken by themselves, Gemini/Chat/Copilot/etc. and a long list of CustomGPTs. The educational process will be crowded, a far cry from the old teacher-student model some believe still defines the educational process.
Accreditors and credentialists must immediately rewrite the rules as GenAI becomes a major player in the educational process. It’s a GenAI/faculty/student threesome now – and forever. Watch as the differences among correspondence, distance learning and online programs shrink to nothing – and the expanding role GenAI will play in the process as the regulators of correspondence, distance learning and online programs struggle with the merger. The same process will play out in the classroom where roles will also be re-defined, re-invented and re-imagined.
At the end of the day, it’s an exercise in the reallocation of educational power, accountability, standards and appropriate compensation.
Key? The definition of “academic engagement” will have to change as the role of GenAI/CustomGPTs explodes. Just look at five parts of academic engagement as defined by the Code of Federal Regulations:
“Attending a synchronous class, lecture, recitation, or field or laboratory activity, physically or online, where there is an opportunity for interaction between the instructor and students …
Submitting an academic assignment …
Taking an assessment or an exam …
Participating in an interactive tutorial, webinar, or other interactive computer-assisted instruction …
Participating in a study group, group project, or an online discussion that is assigned by the institution; or … interacting with an instructor about academic matters …”
GenAI will forever change the definition of student/machine engagement. It already has.
Now what?
Some universities will just keep their heads in the sand for as long as they can, often with unrealistic punitive policies about GenAI as a growing number of their students use ChatGPT, Gemini and other tools to summarize, convert, write, research and take tests, among many other class tasks.
Other universities – the smart ones – will immediately assemble open-minded Task Forces to explore how and where GenAI will impact higher education. A few innovative ones – like MIT – will immediately pilot aspects of automation to determine what parts of the education process can be in-sourced, co-sourced and outsourced. They will most likely conduct this experiment with online courses where the possibility and taste for automation is the greatest.
Finally, universities should proactively engage with those who accredit educational programs to push them toward defining how “engagement” and other activities should be defined as GenAI moves from assistants to partners and beyond.
link