AI and the Future of Education
Artificial intelligence (AI) has been around since the middle of the last century. Really! It is also about to revolutionize our lives in ways we cannot now conceive. If this sounds like something it is hard to wrap your mind around, you ain’t seen nothin’ yet.
AI allows machines to do things that would, until now, require a human intellect to perform, such as problem-solving, decision-making, reasoning, conversing, and creating intellectual property in the areas of science, literature, art, and music. In its most basic sense, it works by analyzing data to discover patterns. This is accomplished through “machine learning,” which is the application of algorithms that permit the programs to improve their ability to simulate human intelligence–in other words, to improve with practice!
AI as developed today dates back at least to 1948 when Alan Turing developed a program called Turochamp that could play an entire game of chess, although at nowhere near the level of a grand master. Even so, the program never went into production because the relatively rudimentary computers of that time were incapable of running a program of that complexity. It took merely another 68 years for DeepMind to devise the AlphaGo program, which was able to defeat Lee Sedol, an 18-time world champion player of Go, a far more complex game than chess: there are more possible board positions than there are atoms in the observable universe.
In May of this year, just 10 years later, four large language models, i.e., programs that are meant to simulate language, were able to pass the Turing test (yes, that Turing)—they imitated human speech so well that the human subjects in the experiment could not tell whether they were conversing with a human or a machine. Not only are machines getting much closer to replacing humans, but they are getting better at it faster than ever before. Can it keep up?
Maybe. But there are obstacles that could just as easily cause the progress in AI to pause. Indeed, the operative term is “artificial.” Until what is thought of as “common sense” can be reduced to a data point and fed into a database, a computer will not recognize it. Humans do not learn by digitizing information. As long as that body of knowledge-based intellectual functions cannot be defined more precisely than as common sense, AI will remain just that–artificial. While computers are already better than humans at reading X-rays and MRI scans, your PC will still tell you it is more efficient to walk to a nearby gas station to fuel your vehicle than to drive there.
There is also the matter of infrastructure. As ephemeral and intangible as all this cloud and web talk may seem, AI demands a great deal of computing power, which in turn demands a great deal of electricity, an amount that the grid in its current configuration will not be able to supply.
Five big companies known as hyperscalers–Oracle, Alphabet, Microsoft, Amazon and Meta–have been borrowing heavily to invest in data centers. Their combined projected investment for 2027 is $1.2 trillion, which would exceed the current United States military budget by $50 billion. Profits that turn out to be lower than predicted and unanticipated borrowing costs could serve to curtail some of these investments and slow the progress in this field.
Moreover, data centers and electric utilities have been receiving some unexpected pushback. In 2025, electric utilities received data center power connection development requests for over 700 gigawatts, almost double the amount of all electricity consumed in the United States only two years previously. As a result, utilities have asked for a combined increase of over $29 billion for the first half of 2025. Over 30 states have introduced bills concerning data centers in an attempt to address both cost and environmental concerns. New Jersey bills S.4143 and A.5564 introduced in 2025 would require that all electricity be derived from clean energy sources.
Another check on the advance of AI may be the very human tendency to procrastinate. Electricity was far more efficient than steam power, yet it took more than 30 years for the technology available by the late 19th century to be widely adopted in manufacturing.
And yet, whatever we may feel about data centers, and there seem to be no voices other than the technology companies themselves that display anything other than abhorrence for them, are we ready to give up Google? What about those summaries that Google now supplies before directing you to a website with the answer to your query? In Penske Media Corp. v. Google and Alphabet, the publisher of Rolling Stone, Variety and Billboard has brought suit in the U.S. District Court for the District of Columbia in which it accused Google of eroding traffic to its websites (theretofore directed thither by Google’s algorithm) by adding AI summaries to search results. The search itself, of course, is performed by AI that is so ubiquitous as to be simply a part of the landscape. It is unlikely to take another 30 years to adopt AI wholesale.
How then does it fit into a 17-year education career (not including preschool)? In order to answer that, we need to dig down deeper into a question that we have not had to consider since we last wrote an essay for a college application: What is education for? Is it to prepare students for particular jobs? Sadly, for many graduating seniors, the computer coding jobs for which they opted to prepare themselves when entering college have now, over the period of those four years, become obsolete. Computer coding is now done by AI. What will an AI-saturated employment outlook be in another four years, or even eight or 12 or 17?
If the needs for a human work force can change so quickly, perhaps any vocational training is short sighted. What, then, do we need in order to engage with whatever world the next cohort of graduates will face when their initial formal education comes to an end? Data center protests, lawsuits, and infrastructure costs notwithstanding, AI is here to stay and all students will need to deal with it.
At the very least, they will need to learn to read and to think critically and to evaluate their interactions with AI. They will need to be thoroughly computer literate and to be able to handle volumes of information orders of magnitude larger than anything we have seen thus far. Graft all of this onto a standard 20th-century curriculum and the problem becomes obvious: what to trim.
There has been a faint whiff of complaint wafting through the zeitgeist that cursive is no longer taught. In light of the fact that there is an opportunity cost to everything, should it be? Should writing be taught at all? What about grammar, foreign languages, chemistry? All of these as well as other subjects once thought to be necessary to the development of an educated citizen may well find themselves on the chopping block as time goes on.
Once the content of education has been decided, how then will it be imparted? An approach set forth as recently as February 2025 in Frontiers in Psychology reads almost as quaint. It speaks of increased efficiency, time-saving, and access to more resources on the one hand and digital fatigue, loss of interpersonal skills and isolation on the other, as though Google and Facebook comprised the AI universe, when they are barely comparable to the invention of movable type in the development of universal literacy.
Much of schoolwork that is now done by students, such as writing essays, written homework, performing literary research, can now be done entirely by AI and even more can be done as they progress up the academic ladder. And will there be a need for a professor to hold office hours when an accommodating AI bot can be reached at any time? Or a need for a professor to be there at all? A return to chavrusa learning where partners read a text aloud and then discuss–and argue–its meaning might at first blush seem to obviate the difficulties posed by AI, but even here AI could be used to substitute for a partner or even a teacher.
To the question, “Are human teachers passé?” the answer is “Perhaps soon, but not yet.” In our current real world, common sense still counts for something. The internet, for all its breadth of knowledge, has still not cottoned to the fact that I do not own a dog.
Inexpensive and Free Online Courses for the AI Neophyte
Generative AI for Everyone by DeepLearning.AI
www.coursera.org
Elements of AI by University of Helsinki
Artificial Intelligence for Beginners
microsoft.github.io/AI -For-Beginners/
Sue Kleinberg is a contributing writer to Jlife magazine.
