The Future of AI in Higher Education
Artificial Intelligence (AI) is here, so what does the future of AI in higher education look like? The following is a projection of where I envision higher education heading as it adopts more AI technologies. This projection is based on the current state of higher education, the rate of current advancements, and slight assumptions about how the general population will adopt these technologies.
First, let’s all catch up on what AI actually is. Artificial Intelligence, as we currently know it, is a sub-product called a Large Language Model, or LLM. An LLM is trained on vast amounts of language data that allow it to infer, understand, generate, comprehend, deduce, and contextualize the language on which it was trained. It is further trained on specific or general data, depending on its purpose. Examples of LLMs are Microsoft’s Co-Pilot, OpenAI’s ChatGPT, and Anthropic’s Claude. LLMs allow users to do what only computer coders could do previously. LLMs allow us to code programs using human language, and this is how LLMs became so widespread and why AI is advancing so quickly. We are only scratching the surface as more people learn how to adopt LLMs into their workflows and daily lives.
So, what does this mean for higher education, or just education in general? As we’ve seen with TV/movie streaming or the expansion of food delivery applications, people want more control over their time and how they interact with the world around them. They want to watch what they want, when they want, and would rather spend money on food delivery so they can spend more time doing other things. People want increasingly more control over their time and interactions with the world.
Let’s now take this thinking to higher education and how AI technologies can impact it. AI will soon reach a point where an agentic AI bot can be created and trained specifically for an individual. I imagine the business model won’t be too different from how your average subscription model may work. You pay a monthly fee for access to your own agentic AI from a provider of your choosing. But over time, this AI agent will learn about you—your favorite music or movies; it will learn your favorite restaurants and how long it takes you to drive to them; it will learn who you spend your time with and what you like to do in social settings; and, among many other things, it will learn how you optimally absorb and retain information. And this is the key to the fundamental change in higher education.
Students at all levels of education will soon have the ability to bring an agentic AI along with them through life. The longer they utilize it, the more effective it will become. If a student uses an AI agent starting in third grade, then by the time they reach college, their agent will have an incredible understanding of their ability to learn. Moral and privacy concerns aside, its goal is to optimize your life and help you as best it can with the information it has.
No longer will instructors be the forward-facing entity in the classroom. Each student will be able to access information and learn in whichever way suits them best. Maybe it’s a series of flash cards or a long study guide—possibly a hybrid of the two. Maybe it’s in the format of a game show or a podcast. It could be a video game that weaves concepts into the gameplay. The point is that it is entirely up to the student’s AI agent to use the instructor’s input (worksheets, books, quizzes, figures, etc.) and output it in whatever format is needed for learning.
No, this doesn’t mean that instructors or professors are less valuable. I would argue the opposite. They are the foundation, the back end of this multifaceted educational machine that can share knowledge in limitless ways. Instructors or professors collect the concepts, organize the course, and still use their knowledge and expertise to ensure the overall message and ideas can be communicated the way they intend.
Sure, there could still be large lecture halls comprised of hundreds of people, but the professor or instructor no longer must be the entity providing the information. It could be an AI avatar with a cloned voice of the instructor or something else entirely, creating time for the instructor to work on something they previously never had time for. At a larger scale, instructors could be teaching the same course at the same time in multiple places. Maybe this means that institutions can more easily expand around the world. Now they have more tuition coming in and can close that decrepit building that’s been on the books for far too long. This opens brand-new possibilities in higher education to ensure that, no matter the institution, students are receiving the best possible education using the faculty member’s intellectual property (IP) that they have created.
IP will play a significant role in higher education. Let’s think through an example of how IP could be used in a higher education industry that fully adopts AI: If a professor at the University of Iowa were teaching a class and expressing concepts, organizing lessons, and creating documents that allow students to understand the material far better, even when using an agentic AI to help, then other institutions would want to provide a similar class to Iowa’s. They could try to recruit that faculty member, but if that doesn’t work, they would still look at other ways to supply their students with the best material possible, as the gap between average and superior instructors may become more visible to students. Without approval from the faculty member, using their IP would not be possible. Institutions would have their faculty members’ IP safely secured within their ecosystems so long as they are employed there. However, if the University of Southern California (USC) wants to license that IP for a fee, it would now be possible for that professor to provide their lessons, documents, and concepts to teach at another institution. A faculty member at USC would facilitate the course using materials provided by Iowa’s faculty. Iowa’s faculty member would collect a licensing fee from the other institution, and a cut of that would go to their current institution, thus opening a brand-new revenue stream for colleges and universities across the country.
Education is a large, but not the sole, arm of a university. Research and discovery at many universities are massive enterprises that bring in millions of dollars each year in external funding. The onset of AI in higher education will bring tremendous value to these areas.
What happens when AI has absorbed, vetted, and understood every single ounce of knowledge that humans possess? The value of new ideas and discoveries will skyrocket. That means places like large R-1 research institutions will be on the leading edge of providing new knowledge and informing the latest and greatest AIs in the world. There will be a much larger spotlight on research and discovery than ever before, and the arms race to provide the newest breakthrough and latest piece of knowledge will be more apparent than ever.
What about institutional websites? If agentic AIs are scouring the internet and providing relevant information to their users, does that count as a web visit? We can assume so, but the value of web traffic will plummet since web traffic will almost solely be comprised of AI. How can institutions better provide information to their audiences if they are receiving most of their content through AI facilitators? What does this mean for various web teams at universities and colleges when most web traffic is not human? These are some questions that higher education institutions will have to discuss.
The question is no longer whether this is going to affect higher education; it is now when. These LLMs are already out there, and the proverbial genie is out of the bottle. It is up to the leaders at higher education institutions to understand what this means and prepare for this rapid change, which may be coming sooner than we think.