The Glass Horse: Discernment and Artificial Intelligence
On the press tour for his new film The Odyssey, Christopher Nolan commented on the rise of artificial intelligence technology and the skepticism around it: “AI is a Trojan Horse that everyone knows the Greeks are inside… it’s a transparent horse, it’s made of glass, and everyone can see what’s inside it.” He called the pushback against AI, especially among young people, a “healthy skepticism.” Summing up, he said, “Technology is always going to give us great gifts… but it has to be viewed with skepticism.”
Nolan has never been one to rush away from innovative and exceptional technologies. In fact, he built the shooting of The Odyssey around IMAX cameras. Notice that he doesn’t say that AI is not a gift at all—but he supports the people who don’t simply take it at face value.
When we consider how new technologies can support education, Lexandria takes a similar approach to Nolan’s analogy. AI is a gift. It’s also one we have to use cautiously. On the one side, there are people who have become neo-Luddites, hoping AI will simply go away. On the other, there are people who trust generative models over human wisdom. We can walk the narrow path of using AI well, rather than leap off the cliff on either side.
Discernment is the key: not just in education, but in life. A good education is not mere recitation of facts and figures, but gaining wisdom. Large language models have already surpassed humans’ ability to store data. We cannot take that as a sign that we should turn over our ability to develop tastes, values, and good judgment.
Why the Gate’s Still Open
The Greeks are visible, but the Trojans still open the gates anyway. That’s the implication of Nolan’s analogy, even though he chose to focus on the skepticism that young people feel toward AI. Many people seem to think that AI is pure “slop,” that AI users are motivated solely by poor taste, or that AI’s advance is solely the result of conspiracies by the wealthy. There is talk among anti-AI circles online of an “AI bubble.” Those beliefs are a sign of overly motivated thinking: Assuming that the only way for someone to disagree with you is for them to be deficient or deceived somehow.
The glass horse does not need to be a grand conspiracy to enter the city. It only needs to offer something better than what came before. And AI does: It offers convenience and capability that was once reserved for a narrow slice of society, now available for free to try. The median American can afford a monthly subscription to the major AI providers for less than an hour’s work. The product speaks for itself. People with busy lives, little time or inclination to develop complex skills like coding or art, can now see their ideas brought to life in minutes. The people who want to see that convenience and capability go away will be sorely disappointed.
People have too little time to do everything they want to do. That has always been true. And for generations, we have lived in a time of technological development that focused on saving physical effort. The washing machine, for example, has alone saved untold millions of people hours of labor—so successful, so ubiquitous, that it is difficult for a modern person to comprehend just how much time it used to take to do such a seemingly simple task.
Then technology began to turn toward intellectual advancements. An abacus was the standard for millennia before calculators could be invented, and graphing calculators themselves transformed the way that mathematics could be taught and learned. Calculators did not replace the student’s need to understand numbers and formulas.
Likewise, new tools do not replace the student’s need to understand parts of speech or how to construct evidence.
The Horse Is Not Alive
It is unfortunate that the blanket term “artificial intelligence” is applied everywhere. Complexity is at an unprecedented scale, but a large language model is more like a graphing calculator than a droid from Star Wars. R2-D2 and C-3PO might be the unjust property of humankind in a galaxy far, far away, but Claude and ChatGPT are computer programs.
That means that the human element will never be replaced.
At Lexandria, we use AI tools regularly. They can generate research leads, help initiate product elements, and stress-test an argument or a lesson plan before it goes out. But there’s a reason the model never gets a byline. We will never blindly copy and paste the outcome of a tool’s best guess. A human being has to check sources, evaluate arguments, make value judgments, and take responsibility for what comes next.
In the classroom, teachers can follow a similar distinction. AI tools can help you refine your preparation for class, but they will not stand in front of your students and answer their questions. A teacher can drop discussion questions into an LLM for a stress test, but managing the discussion is entirely a human experience. Judgment calls on what AI can and can’t do are essential to understand, because that’s what students will have to do—and often.
A student who cheats on a paper by copying verbatim what an LLM said isn’t hurting the teacher, but himself. The student who skimmed SparkNotes instead of doing the reading is only making it harder on herself the next time her teacher asks questions. Labor can be delegated, but judgment cannot. That holds true whether it’s an LLM doing the delegated labor or a human.
That’s why Lexandria’s library isn’t a collection of AI summaries. We want teachers to have AI-powered tools, but to retain the judgment of the full library. Our teachers aren’t having AI summarize Democracy in America for their students. You cannot shortcut the actual work of education. Think of the washing machine: When we do use AI, we’re saving our hands from scrubbing, but the clothes still need to be put in, the machine started, the clothes worn by real people, for any of it to mean anything. A student can benefit from their teacher using AI-assisted tools like our standards mapping search function. They still need to learn how to judge a high-quality analysis of Alexis de Tocqueville from a poor one.
Show Your Work
AI summaries can tell students what a plausible answer might sound like. They cannot replace the experience of making judgments. An AI summary or an AI-drafted response is a facsimile of judgment. Using the documents in the classroom-forward way that Lexandria encourages will allow students to encounter reality and judge between plausible answers.
There are myriad ways that we can use AI to aid, not damage, our experience. Christopher Nolan and other filmmakers are having to contend with AI as a haphazard tool, a creator of artistic expressions at the push of a button. This is why so many people see “AI slop” everywhere: democratizing visual media does not magically grant good taste to its users.
Civics, especially at a high school level, might be experiencing the proliferation of AI in the opposite direction. If AI can perform a task indistinguishably from a human (at least in the teacher’s experience), then that task may not be the best measure of what civics means. We must refine our approach.
Mathematics learned to live with calculators by asking students to show their work. Social studies now needs its own version of “show your work.” Put the document on the page. Have classroom conversations. Ask what another reader might say instead. AI can create opportunities for those discussions, but it cannot replace them.
At the judgment-first level of social studies education, the question is not whether AI could be used to answer an assignment. The question is whether it was a tool to aid judgment, or if it was used as a facsimile of judgment.
The process of the Declaration of Independence’s writing is a perfect example of judgment at play. Thomas Jefferson did not write the document in a “closed book test.” He pulled liberally from existing documents. The final draft did not simply look like his first draft. AI could have, in theory, replicated the first step: Drawing from what existed to make a plausible-sounding draft. It could not make the judgments Jefferson or his fellow revolutionaries made as the process continued.
Civics education should give students some of that same experience. It should create the same willingness to evaluate, to question, and to make judgments. The original documents, therefore, have to come first. AI can plug in definitions. It cannot convince a young student that she should care about the experience of judgment. That is why a teacher’s role will always matter. With a firm foundation in the facts and encouraged by a good teacher, students can develop the skills to make good judgments—which is what civics education is all about.
Christopher Nolan is right to be encouraged that young people are skeptical. Skepticism is not, itself, discernment, but discernment begins when we stop to ask questions. It helps us look past the polished answers and learn what’s beneath the surface. In itself, being skeptical of “AI slop” might be the first step to a new generation that understands the deeper meaning of education not as simple memorization alone, but as the foundation of something greater.
