University budgets across the country are broken. Overall revenue and accumulated financial support appear to be declining for a wide variety of reasons.
New funding policies, administrative reorganizations such as those at the Department of Education, lean fiscal times for states, diminished regard for higher learning, fewer requirements for degrees among employers hiring for entry-level positions and the impact of artificial intelligence all come together to reduce the pool of new students, tuition revenues and grants. As a result, new initiatives are stifled unless they show promise to immediately reduce costs and generate new revenue.
The cost of developing, designing and teaching classes is often largely determined by the faculty and staff costs. Long-running lower-division classes at some universities may be taught by supervised teaching assistants or adjunct faculty whose salaries are lower than tenure-track faculty’s. However, we are now confronted with highly capable technologies that require little to no additional investment and can bring immediate revenue positive opportunities. Each university very soon will have to determine to what extent AI will be permitted to design and deliver classes, and under what oversight and supervision.
However, few of us in higher ed seem to realize that such technologies are freely available today. The tsunami of new and improving AI technologies has inundated us over the past three years so quickly that many in academe have not been able to keep up.
Case in point is the rapidly expanded and enhanced AI app. We know them by the names Gemini, ChatGPT, Claude and a few others that we tried out in 2023 or 2024. Early on, they were deficient in many ways, including hallucinations and a strictly limited number of words you could put into a prompt. That has largely changed, though hallucinations can still happen—which is why it is good practice to compare prompts submitted at multiple sites, as has been suggested previously in this column.
AI is now capable of teaching higher learners in a highly sophisticated way. Given a thoughtful, detailed prompt, it is able to implement most of the proven pedagogical approaches, many of which are unknown to our teaching faculty. It is able to dig more deeply into a topic with more recent data than many of the faculty members are using. It can interpolate, extrapolate and pursue conjectures to their statistical conclusions, revealing unexpected outcomes. It can be friendly, supportive, patient and also challenging at the appropriate times.
My point is that the prompt today can be a most powerful teaching tool that can consistently create a robust, engaging learning environment. It can even offer material through a variety of pedagogies. A well-written, detailed prompt can be the equal of many of our teaching assistants, adjunct faculty and, yes, full-time faculty members who have not been deeply trained in effective pedagogy and current practice.
Let’s test out one of these technologies right now!
This one below, shared by the “There’s an AI for That” newsletter, provides an online tutor to teach students using an assortment of time-tested technologies and pedagogies. The first step begins with merely a prompt, albeit not a short one. The Interactive Learning Tutor is a virtual instructor that runs on the leading frontier AI app models (Gemini, ChatGPT and Claude). Most interesting is not only the vast range of topics it is prepared to deliver, but the range of pedagogies it will enable at the request of the learner. So, the learner can pick their pedagogy rather than being subject to the designer’s and instructor’s choice of teaching and learning methods.
“This prompt turns AI into a dual-role system that first helps learners choose the most effective study techniques for their subject, then switches into live teaching using those chosen methods. In Navigator Mode, the system gathers details about what the user wants to learn and how they prefer to learn it, then presents a curated set of three to five evidence-based methods from its catalog. Each method is explained in depth, with strengths, limitations, step-by-step applications and concrete examples tied to the user’s subject. It also suggests learning archetypes, outlines pitfalls and fixes, creates an integrated action plan, and provides tools like reflection prompts, comparison tables, and pathways matrices to help learners see their options clearly.”
This is a highly qualified virtual teaching assistant, eager and well prepared to take on the task of teaching an individual in an online tutorial or class. Imagine 30 students, with a week or so initial orientation, running this app to achieve specified learning outcomes in an engaging, interactive way. The materials provided in the individual tutorial could be augmented through a syllabus of supplementary readings, reference materials and periodic assessments leading up to a final project or paper to be submitted individually or by groups in the class. The prompt and app take over the most important role of engaging and guiding the student in ways proven to be successful.
So, let’s give this a road test! Step one is to copy the 1,112-word prompt that begins with and ends with . Paste this into the input box of your chosen AI app and press enter. After a brief moment for processing, the app will reply with an invitation to share the topic about which you would like to learn. This would be defined by the instructor of the course along with the learning outcomes for the class. As an enthusiastic learner who is going to attend an outdoors bluegrass music festival today, I typed in, “I want to learn about bluegrass music, including how it evolved, its popularity, the geographic range of its performance, the instruments used and cultural impact.” Including such learning outcomes will streamline the process for later assessments.
Gemini replied, “That’s a fantastic topic! Bluegrass music has a rich history and a vibrant culture. We can definitely break down its evolution, instruments, and cultural impact.” Gemini went on to write,
Now, to help me suggest the best learning methods for you, could you tell me a bit about how you prefer to learn? For example, when you’re trying to understand something new, do you find it helps to:
- Discuss it and ask lots of questions?
- Organize information visually, maybe with timelines or diagrams?
- Listen to examples (like different songs) and connect them back to the concepts?
- Try to explain it to someone else to make sure you’ve grasped it?
Gemini also offered a personalized action plan to set the schedule.
Integrated Action Plan
- Immediate Actions (First 1-2 Hours): Use the Socratic method to generate a list of foundational questions about bluegrass. Use dual coding to create a simple timeline of major events (e.g., Bill Monroe forms the Blue Grass Boys, Earl Scruggs joins, the Newport Folk Festival).
- Medium-Term Actions (First Week): Use the Storytelling method to weave those timeline events into a narrative about the birth of the genre. Pick one specific concept (e.g., “the high lonesome sound”) and use the Feynman technique to explain it in your own words.
- Long-Term Actions (Ongoing): Continue to build out your visual library (dual coding) with instrument diagrams and regional maps. As you learn more, continuously refine your core story (Storytelling) and test your understanding of complex subtopics (Feynman technique).
Further details were shared by Gemini in table format to ensure that the learner understood the different approaches. Then it gave me, the learner, the chance to affirm that I wanted to blend those modes. And then we were off and running.
This format seems ideal to prototype just-in-time certificate programs, over-demand programs where insufficient numbers of faculty are readily available and other instances where time or finances may not permit full design and teaching staffing. Carefully monitor each student by asking them to share their exchanges with the app. Be vigilant to protect the learner’s FERPA rights. Use secure methods for quizzing and other assessments. Compare the outcomes to other such classes and gather feedback from the participating learners. Modify the original prompt to meet your needs and refine the results for the next offering.
