– Alex Cortez, September 2026.
Like everyone, I am trying to figure out AI’s role in my life and my work. Will it transform my work? Will it put me out of work? Will it enlighten me or will it encumber my thinking? Will it augment or will it atrophy my skills in alliteration? Will I end up with a bad case of FOMAI (Fear of Missing AI)?
As with many things in education (and the world more generally), I find a lack of common language around AI a challenge. In the last year, I have heard so many discussions – excited, fearful, and often both – about how education organizations are approaching AI, and I think there are at least six use cases for AI in education being considered by students, families, educators, funders, and policymakers:
1. How we teach students
2. What we teach students
3. How we assess students
4. How we advise and support students
5. How education organizations operate
Before I “delve” into each of these, a few observations on AI that have shaped my thinking (and which AI has told me make this essay feel like two essays – good to know):
Those observations are the lenses through which I’ve approached AI in general, and specifically in considering the following six use cases of AI in education:
How we teach students. AI moves the prior decade’s promise of personalized learning from hype to hard reality (I really do love alliteration). AI offers essentially infinite ways to customize content and context for each individual and community.1
Imagine every lesson is taught with dynamic content that combines the scientific intellect of Albert Einstein, the wisdom of James Baldwin and Audre Lorde, the historical insight of Ken Burns, the comedic timing of Chris Rock and Robin Williams, the orchestration of John Williams, the direction of Kathryn Bigelow, and the production values of a James Cameron film. Then imagine that every learner can change that “playlist” to whichever individuals most speak to them.
A standard critique is that AI won’t motivate learners the way a human can. This feels a bit disingenuous because plenty of teaching is currently done by people not particularly motivated to motivate their students. Students are motivated by different things and in different ways. AI provides the potential to try many methods and, if structured right, to hand individuals more agency to find what most motivates them.
AI can also transform teaching by unbundling different teaching tasks. Not every educator is strong at lesson planning, lecturing, leading group discussions, advising, or grading. With AI, some of these roles can be disaggregated and made easier, hopefully enabling teachers to focus on what they have the aptitude and appetite to do, and freeing up as much precious human time as possible to engage with students.
Finally, AI can enable learning to be both competency-based (determining if a level of learning has been achieved) and competency-paced (enabling individuals to achieve competency on their own timeline).
What we teach students. AI is a new tool, and it will reveal new knowledge and skills to learn – like how to use AI. It will also alleviate the burden of learning some existing knowledge and skills that will atrophy or otherwise be outsourced. This did not start with AI. I am sure that without Excel, my math skills on their own would be pre-algebra. With Excel, I can model reality. But because of Excel, my old HP12C financial calculator is now detritus sitting in a drawer, and I have never had to learn how to use a slide rule (and some people may now be asking AI, “What is a slide rule?” or “What is an HP12C?”).
Reading this, one reaction may be: “but you still must have the durable skills for things like framing a problem, critical thinking, communications, and facilitation to know what analysis to do in Excel and how to make it valuable! These skills are becoming even more critical!”
That is correct, but I would put the emphasis on “still.” These durable skills have never not been important – they have just often been poorly taught to most people. However, with the outsourcing of some skills to AI and the continually shrinking shelf life of some technical skills (even as new ones are born), these durable skills are the ways to remain relevant and competitive. They also tend to make people better at interacting with other human beings – never a bad thing to encourage.
How we assess students. The concepts of adaptive testing, personalization, and teachers using assessments to be nimble and responsive to students and their classrooms are not new. However, AI is a tool that can supercharge and automate these efforts for educators by collecting vastly more data, analyzing it almost instantly, and suggesting actions.
AI can also be a tool to assess and develop teachers. AI can gather and analyze real-time data on classroom instruction and formative assessments to equip teachers with daily insights to shape their teaching.
Any discussion of assessments invariably includes a discussion of cheating. Cheating on assessments has likely been around for as long as there has been something worth assessing. AI is a tool for teaching and testing – so it is naturally a tool for subverting teaching and testing.
Perhaps ironically, the biggest impact AI will have on assessment is creating more offline assessments: proctored assessments on paper or on heavily curated and closed systems (e.g., laptops with no internet access), oral exams, and/or experiential demonstrations (be it a project, a clinical rotation, a performance, or success in building something).
It is tempting for us to throw up our hands and simply say “the only people students are cheating are themselves!” but that fatalism has a few flaws. Education serves many roles. One is to signal an individual’s value in the marketplace for things like employment and graduate school. Cheating undermines this. While we don’t always like to admit this, education in our society is also about competition. If anyone can gain an advantage from cheating, everyone will feel the pressure to do so. I’m not letting AI off the hook for making it easier to cheat – but it didn’t create the hook. Beating AI on cheating is possible – it’s just very inconvenient, and may advantage people with good presentation skills or handwriting.
How we advise and support students. There is no shortage of student needs – tutoring, academic advising, college and career coaching, mental health, navigating complex external processes and bureaucracies, etc. We’ve tried to systematize these supports for decades to improve quality and give the people delivering them more leverage: first paper curricula and training, then software platforms, then chatbots, and now AI.
All of this has been in service of three important goals: (a) to scale to reach more students in need, (b) to succeed in helping them achieve the outcomes they want, and (c) to do so in a way that is financially sustainable.
However, as noted earlier, quality human supports are finite and expensive. Unless we create a profound societal shift in how we support human flourishing through other humans, and secure the funding to sustain it, scaling will continue to be limited. Advising organizations are feeling pressure / seizing the opportunity to adopt AI because it provides a way to scale quickly. I think the jury is still out on whether this remains a long-term structural trade-off between breadth and depth of impact.
One worry is that AI will be a biased adviser because it is trained on the internet, which has strains of racism, sexism, and general meanness (because humanity has those strains).
That problem may be solvable by properly curating the information AI advisers draw from, but if we do that well, we might then need to contend with the idea that AI can be LESS biased than humans. Of course, we want to eliminate insidious negative biases. However, we also have to recognize that love is biased, and the counsel of those who love a person is shaped by their own lives, their worldviews, and their hopes for the person being counseled. I think this is where AI, if properly constructed, can distinguish itself. We want any advising, human or AI, to inform, inspire, and influence… but not impose. This tension between acting as a caring sibling and acting as Big Brother existed long before AI.
Like students, teachers benefit from coaching, and people who can coach teachers well are always in short supply. Just like student advising, teacher coaching will have to wrestle with when AI amplifies coaches, when it replaces them, and when it simply equips them to coach better.
How education organizations operate. Organizations are incredibly complex… and usually not in a way that promotes highly satisfying functionality. AI can be used on just about any recurring process to reduce variability, improve overall quality, and increase efficiency that buys back time, resources, and – if customer-facing – a healthy amount of human goodwill.
In education, it could be used to automate grading or test generation. AI can help systems better coordinate – like the registrar and bursar at a college working together to improve the student experience. It can also be used to address organizational process breakdowns. Years ago, I interviewed the president of a community college who discovered that they were losing students who dropped out rather than taking the medical leave they actually wanted because the college required completing 12 steps across different departments – many of which didn’t know the others existed. This is solvable without AI – but AI deployed in improved processes can solve it faster and at scale.
However, all of this could be disruptive to education organizations and the people they employ. We worry about what AI could do to fields like accounting and law – but what about fundraising when AI can automate grant proposals, grant management, and grant reporting, and philanthropy uses AI to review them?
The information and insights available to execute 1 through 5. I am still an AI neophyte, but it already feels like AI functionality has become a commodity that is accessible to almost anyone. It took ChatGPT 14 minutes and 36 seconds to build me an Excel-based customer relationship management (CRM) platform, including instructions for how to sync my emails to automatically download to it through some creative re-routing (take heed, Salesforce). I also came to realize that this was complete overkill for my needs – I can just have AI scour my files to do the equivalent.
In the long term, what is going to most distinguish the capacity of education organizations and systems to create transformational value with AI is the power of the proprietary data that they use to fuel AI functionality.
In the postsecondary advising space, there is a lot of public data that government systems should make available to the public, but don’t. When you hoard data, you hoard power. Fortunately, a number of postsecondary advising organizations do the hard work of collecting data from students. They are building powerful and proprietary data sets that they then share with those they advise and/or other advisers they support, and those data sets become more useful every year.
To return to the college medical leave example in #5, AI can be used to analyze where students are encountering problems and not only help solve them in the moment but also identify changes in policy and practice that would address the problem in perpetuity. AI can also use data to identify warning signs that could prompt support for students who are at risk before they take a medical leave.
One additional danger to consider comes not from AI, but from AI companies. I have heard from nonprofit leaders that they are facing increasing pressure to give their proprietary data to AI companies without continued ownership or compensation for it. This is like training AI to replace you, only it won’t do as good a job because it shuts out one of the key sources of new data – humans interacting with students.
Conclusion – AI is a Force we must reckon with in the physical world as well as the virtual one. I sometimes wonder if the biggest risk of AI is not on a screen but in AI-driven physical work. I suspect that many of us are underestimating just how disruptive AI will be when given corporeal form.
For example, in 2025, Amazon’s Vulcan robot demonstrated it had mastered all three human grips (pinch, power, and precision) and can pick and stow 75% of warehouse items at human speed. Amazon optimistically noted that deploying Vulcan has “created hundreds of new categories of jobs at Amazon, from robotic floor monitors to on-site reliability maintenance engineers.”2 Still, an October 2025 New York Times article cited leaked internal documents indicating that Amazon plans to automate 75% of its warehouse operations, avoiding the need to hire 600,000 employees by 2033 (though this does not necessarily mean firing existing ones – but who knows).3
Vulcan: live long, but who prospers?
That does bring us back to education. Education provides us with the means to achieve agency and exercise power – including the power to deploy or constrain AI.
In education and elsewhere, AI perhaps started as a lightsaber in Star Wars – a tool that is easy to pick up and wield. However, AI is quickly becoming the Force – a powerful energy field that “surrounds us, penetrates us, and binds the galaxy together.”4 The Force is harder to master, but ultimately more valuable.
And like the Force, AI has a light side and a dark side. The light side is an AI that extends human capacity. It gives a tutor more leverage, an adviser better information, a community college student a path back from the brink of dropping out. It gives students more knowledge, skills, and agency. The dark side is an AI that displaces humans rather than extending their potential, and concentrates power in the hands of a few to the detriment of the many. In the lore of these movies, the Force does not choose between the light side and the dark side. People do.
Education determines who those people will be and what path they will pursue.