Beyond a keynote speaker: A strategic roadmap for AI in higher education

Key points:

• AI is transformational, but it is not a magic solution that automates the struggle of learning
• The pedagogy gap: Redefining the role of faculty and AI in higher education
• Colleges are adopting AI faster than they can govern it
• For more on higher education and AI, visit eCN’s AI in Education hub

The atmosphere inside the Global Connect/American Accounting Association conference at Caesar’s Palace in August 2026 was a study in contrasts. Outside, the Las Vegas sun was punishing, an environment so relentless that I joked that the swag bags should have swapped out sunblock for flame retardant. Inside, however, the climate was controlled, though a different kind of heat was beginning to rise. This was the figurative fire of artificial intelligence, a technological shift that has left many faculty members feeling as though they are standing in the middle of a desert without a map.

As a professor who has navigated frontier models since the introduction in 2023, I recognize that my role has shifted from a mere instructor to a sherpa or an architect, guiding others through a landscape where information asymmetry is being obliterated. As the opening keynote speaker and to ground the audience before tackling this disruption, I shared an anecdote. I happen to reside just less than 30 minutes away, but I couldn’t resist staying at the hotel anyway, and on the concierge floor last night. You know why? “Those towels. They are so thick and luxurious. I could hardly get my suitcase closed this morning.” Some deadpan silliness completely put the audience at ease as we all roared with laughter.

It served a strategic purpose. In the world of digital pedagogy, humor is a vital tool for reducing transactional distance, which is the psychological and communicative gap between the instructor and the learner. By opening with a shared laugh, we lowered the collective anxiety of the audience, preparing the group for a session of group therapy for the modern professor.

I stand before you as someone who understands the competitive forces at play in our industry. As we transition from the physical heat of the Nevada desert to the psychological heat of rapid technological obsolescence, we must recognize that the fire is already in the classroom. The question is no longer whether we will engage with it, but whether we have the strategic framework to lead our students through the flames before our current curricula turns to ash.

The Wile E. Coyote syndrome: Navigating the pace of obsolescence

The speed of AI development creates a unique psychological burden for educators, a phenomenon I liken to the Wile E. Coyote syndrome. We are perpetually chasing a technological Roadrunner that only seems to gain speed, leaving us in a cloud of dust just as we think we have grasped a new concept. I admit that I have spent nights staring at the ceiling, wondering if the curriculum I prepared yesterday will be obsolete by the time I wake up. This is the Ground Zero state of education today: a world where the frames of knowledge are being redrawn in real time.

For a university leader, this pace is not just a pedagogical hurdle; it is a strategic threat to the perceived value of a four-year degree. For those in fields like accounting, this creates a profound friction. Accounting is fundamentally deterministic. It is a discipline of precision where one plus one must equal two, and the ending balance must tie out with absolute certainty.

Conversely, Large Language Models (LLMs) are probabilistic. They are high-speed guessing machines that predict the next word based on statistical likelihood derived from billions of records. They can be amazingly accurate but do sometimes faulter. When an AI confidently states that Christopher Columbus sailed the ocean blue in 1892, it is not lying in the human sense; it is making a statistically confident guess that happens to be wrong. This non-determinism means that the same prompt can yield different results every time. For a profession built on standardized correctness and tie-outs, this is maddening.

The anxiety of overnight obsolescence is real, but we must realize that while technology moves at a breakneck pace, the foundational principles of human learning remain immutable. We cannot let the statistical guessing of a machine dictate the rigor of our discipline, yet we cannot ignore that the machine is now a permanent fixture in the professional world our students will enter.

Media as vehicles: Grounding AI in instructional design

To find our footing, we must look back to the foundational debates of instructional design, specifically the 1990s discourse between Richard Clark and Charles Kozma. Clark famously argued that “media are mere vehicles” that deliver instruction, suggesting they do not influence student achievement any more than the truck that delivers the groceries. This metaphor is our North Star in the age of AI. The technology (the truck) is secondary to the content and the pedagogical method (the groceries). If our students are not reaching mastery, we must look at the nourishment we are providing rather than simply blaming the delivery vehicle.

AI is challenging our instructional designs because it has exposed the functional obsolescence of many traditional assignments. For decades, we relied on certain containers to measure knowledge: the five-page essay, the take-home problem set, or the research paper. We believed these containers held evidence of critical thinking. However, AI has shattered these containers. If a student can generate a coherent essay or solve a complex problem set by feeding a prompt into a machine, then the assignment was never truly measuring critical thinking; it was measuring a task that a machine can now perform better and faster.

The strategic takeaway for faculty is uncomfortable but necessary: if an assignment can be completed entirely by an AI, it is now likely failing to measure true synthesis and high-level evaluation. The old groceries are spilling out of the truck because the truck has changed, and we must redesign our containers to ensure they still hold value in an era where basic content generation is a commodity.

The graveyard of predictions: Maintaining perspective in the age of AI

History is littered with technologies that were prophesied to replace the teacher. Maintaining a sense of historical humility is essential to avoid modern alarmism. In 1922, Thomas Edison predicted that talking films would replace textbooks and teachers within a decade. He envisioned a world where Physics 101 was delivered by a screen, yet we ended up with the Cineplex for entertainment while the teacher remained central to the classroom. Similarly, in the mid-20th century, television was hailed as the classroom of the future. Programs like Sunrise Semester offered one-way lectures via broadcast, but the lack of feedback and engagement led to abysmal completion rates and the eventual realization that passive consumption is not the same as learning.

The graveyard of technological predictions continued into the digital age. In the 1990s, major publications dismissed the internet as a useless invention, calling it fritter-ware because it was seen merely as a way to fritter time away. They failed to see how it would eventually transform the world and education alike. More recently, one of the largest school districts in the country spent 1.3 billion dollars on an iPad initiative, only to realize within two months that the devices were not aligned with curriculum needs. In 2005, there were claims that students would tweet themselves into mastery, yet social media created more distraction than deep inquiry. We must recall the words of the banker who once told Henry Ford that “the horse is here to stay,” dismissively labeling the automobile a fad. We must avoid that same trap.

AI is transformational, but it is not a magic solution that automates the struggle of learning. We cannot evaluate this technology based on its worst possible use case, just as we do not abandon water because people can drown in it. We must integrate AI into curriculum and must begin from the instructional design of each course.

Watching the donut: AI as the new broadband

The folk singer Burl Ives once sang, “Watch the donut, not the hole.” In our context, the hole represents cheating, the hallucinations, and the fear of replacement. The donut is the positive potential for deep inquiry and enhanced productivity.

I believe we are currently experiencing a Digital Divide Redux, where AI has become the new broadband. Just as we transitioned from screeching dial-up to high-speed fiber in the 1990s, we are moving into an era of high-speed intelligence. If our students do not have access to professional-tier AI tools, they are essentially stuck on intelligence dial-up while the rest of the professional world moves at gigabit speeds. This creates a massive equity risk for universities that ignore the budgetary challenges of providing these tools.

However, we must also be clear about the limitations of current LLMs. I often describe AI as having AI Alzheimer’s, which refers to its limited context memory. It often forgets what you told it ten minutes ago in a long chat thread, leading to inconsistencies that frustrate the uninitiated. Furthermore, we now live in a Zero-Click Reality driven by tools like Google Gemini. When search engines provide an AI summary at the top of the page, it destroys traditional research habits. This zero-click shift is a literacy crisis. If the summary replaces the source, students stop the inquiry process at the first paragraph. This destroys websites that provide deep content and, more importantly, it destroys the student’s ability to engage with nuance.

To illustrate how the media sensationalizes these risks, consider the recent reports about Anthropic. The press treated the situation like a rabid dog running loose in a neighborhood after an AI model accessed corporate sites. In reality, a developer simply left an API door open during testing. The model did exactly what it was programmed to do; it was a human error of oversight, not a sentient machine on a rampage.

Deconstructing the digital native: Literacy and agency

There is a persistent myth that today’s students are digital natives. While they can swipe a phone or navigate social media with surgical precision, they often lack productivity literacy. Most freshman students arrive knowing almost nothing about the professional and productive use of technology. Digital literacy skills are inextricably linked to student academic success. Students are often lost in Excel, unable to construct basic formulas or functions, and are often unaware of the citation and reference tools built into Microsoft Word, as just a few examples.

This gap between social media fluency and professional productivity is where we, as educators, must step in. The new AI Literacy is not a replacement for digital literacy but a vital subset of it. It requires teaching students two critical skills: Corroboration and Ethical Agency. Corroboration is the habit of checking the confident, probabilistic answers of an AI against reliable sources. Ethical Agency is an understanding of when to use a tool to assist thought rather than replace it.

We must remind students of the “168 hours” concept. Elon Musk has the same 168 hours in a week as they do. If Musk can use those hours to build rockets and revolutionize industries, what is the student doing with their 168 hours? Our role as professors is to teach students to own their learning and use these tools to build their own metaphorical rockets, not just to bypass the struggle of a homework assignment.

The architect’s toolkit: Walled gardens and syntopical reading

Our professional role is shifting from the sage on the stage to the architect of learning experiences. One of the most powerful tools in this new toolkit is Gemini Notebook, formerly known as NotebookLM. This tool allows us to create Walled Gardens of content. Instead of letting AI roam the entire internet (where it might drink from contaminated data sources), we can force it to drink only from the water sources we provide, such as our curated PDFs, lecture videos, and trusted datasets.

This enables what Mortimer Adler called Syntopical Reading in his seminal work, “How to Read a Book.” Syntopical reading is the highest form of reading, where one reads across multiple sources to form a unique synthesis. AI can help students navigate these complex sources by integrating multiple sources from which students can read and gather differing information. Also, by providing summaries and answering learner questions in real-time, effectively serving as a cognitive scaffold. It can even transform a dense lecture into a podcast, creating a Professor in a Box that students can listen to while driving or cleaning. This caters to different learning styles (Auditory vs. Kinesthetic) without compromising the rigor of the content. By providing the vehicle, we ensure the groceries actually reach the student, narrowing the transactional distance by providing 24/7 feedback when the human instructor is unavailable.

Case study: The evolution of Captain Excel

To illustrate this shift, let us look at a specific learning problem: the 99% Pilot analogy. In accounting, total accuracy is required. I tell my students that if a pilot lands successfully 99% of the time and today is their 100th flight, you have a problem. For years, I used a self-grading spreadsheet I authored, called Captain Excel to teach complex functions. I demanded 100% accuracy, and it was a highly effective tool until AI made it functionally obsolete almost instantly. Students simply fed the workbook to the AI and received every formula instantly.

Notably, I did not ban the technology; I rebuilt the experience to leverage the new opportunities provided by AI.  Using new AI development tools, which allow faculty with zero technical ability to build functional applications, I created a custom app that functions as a textbook/manual and a practice arena. Crucially, I included a Socratic AI Button. When students are stuck, they do not get the answer; they get a hint or an explanation of the why behind the formula. I made sure to include multiple questions for every function so they must demonstrate mastery through iteration.

The result was transformative. Students told me they had more fun than they ever had in a spreadsheet, and they left the course as Excel experts, not just copy-pasters who knew how to trick a system. This has become my clearest example of how AI can render old designs obsolete while opening the door to something stronger. In the keynote, my examples of AI-driven course redesign, including courses where students design and build software applications, struck a powerful chord with the audience.

The integrity pivot: Moving beyond the integrity police

The strategy of policing AI is a failed one. You cannot be the Integrity Police for 100 students when you only have 168 hours in a week. Instead, we must focus on narrowing the transactional distance through guidance, trust, and agency. The signature of AI writing is often unmistakable: a monotonous rhythm of sentences that are all the same length. It is a bizarre contrast when a student who speaks like Rocky Balboa suddenly writes like William Shakespeare.

I once held an Amnesty Day after discovering a massive wave of AI-assisted cheating on a writing task. The line outside my office looked like a British soccer match, a sea of contrite faces waiting to admit their shortcuts. It was not about punishment; it was a conversation about agency. When students recognize that cheating destabilizes their own intellectual growth, the dynamic changes. However, the future of assessment must also evolve. We may see a return to blue books and oral exams, though these are difficult to scale. A more viable path may be the implementation of local or regional Exit Exams or proctored exam centers with participation within school walls, or with other surrounding institutions.

During the keynote, I shared how faculty can share proctoring exam sessions across many different exam types, with little cost or effort by the school. Schools can create management systems using low code or no code AI to easily build the infrastructure to support curriculum exam needs.  By bringing the final proof of knowledge back into a supervised environment, we can blunt the desire to cheat and allow the rest of the course to focus on the process of learning.

Heterophily and the road ahead

Our mission is to move from Homophily (where birds of a feather flock together) to Heterophily (bringing unlike minds together). In higher education, we often silo ourselves within our departments. However, true innovation in the AI era will come from cross-departmental and interdisciplinary collaboration. We need to talk to the humanities folks, the engineers, the philosophers, and even that cranky old guy with patches on his elbows who has been teaching the same way since 1978. We must bridge these gaps to understand how this technology impacts the human condition. We must embrace the “Wicked Opportunities” described by Robin Lake of Arizona State University. This is the donut, the chance to deepen understanding and use our brains at higher levels.

As the comedian Lily Tomlin once quipped, “We are all in this together by ourselves.” You have the academic freedom to design your own path, but you are not alone in the struggle.

My mandate to you is to start small. Pick one learning problem in your syllabus that is currently broken by AI and solve it with a new tool or a new design. Build a library of success one step at a time. The road to success is under construction, and we have the professional tools to architect a future that is better than the past.