The Brilliance of Student Reflections: Aura, AI, and Medical Care in Historical Context

Eric wrote this after reading and grading roughly 50 essays prepared by Wesleyan University students attending a class he taught on the history of psychiatry during the Winter semester (2025). Students were asked to read Walter Benjamin’s The Work of Art in the Age of Mechanical Reproduction, submit a prompt about the end-of-life care of George Washington and the psychiatric care of King George III to three different AI chatbots, evaluate the responses through Benjamin’s concept of “aura,” and explain — using Charles Rosenberg’s framework — why these 18th-century medical treatments might have “worked.” All student names have been anonymized to protect their privacy.

Reading through my students’ papers for Trouble in Mind: The History of Psychiatry in the United States, I found myself captivated by the threads of insight woven through their work. The assignment had seemed straightforward enough: compare AI-generated responses about the medical care of George Washington and King George III, evaluate them through Walter Benjamin’s concept of “aura,” and use historian Charles Rosenberg’s framework to explain why these historical treatments might have “worked.” Yet what emerged from their essays was anything but simple.

The Puzzle of Aura in Our Digital Age

What struck me first was how students grappled with Benjamin’s notion of aura when confronted with AI-generated text. Julian — whose paper left me marveling — noted that these AI responses perpetuate “a heightened disconnect between the modern user and an ‘othered’ past to which we are both foreign and superior.” This brilliant observation captured how generative AI evokes a sterile impersonality. By withholding crucial context, the machine not only fails to produce a truly comprehensive answer, it also inhibits the user from knowing what they don’t know — thus discouraging deeper research.

Keith took this a step further, writing, “We are estranged from our medium.” He questioned not just the authority of the chatbot’s output, but the entire chain of creation it represents. “Some poor historian has had their work reduced to data points, broken down, and reassembled into something new.” Even when accurate, the chatbot’s summaries are amputated from origin, method, or argument — fragments with no echo. He added: “Humanity can hope. Something must be missing.”

Tessa echoed this critique. “ChatGPT can generate text that sounds completely plausible as a human voice according to grammar and sentence structure but makes no sense in context,” she wrote. Its voice is convincing only if you do not listen too carefully. The moment you do, she explained, the illusion breaks — the response becomes vague, repetitive, devoid of perspective. It cannot build a new argument; it can only repackage existing ones. “Chatbots are restricted to text that has already been created,” she observed, “or text that could reliably be written according to existing code instructions.” Which is to say: the sound of speech without the spark of thought.

Nathan’s reflection took my breath away: “Artificial intelligence epitomizes so much of what defines our modern-day aura. You can create a chatbot, design a game, create a virtual assistant, or generate pretty much whatever you want. Yet, in all three responses, Gemini, ChatGPT, and Claude, there was no essence, no voice, no real perspective — only an answer.” In the margin of his paper I found myself scribbling, “This is so well written, so heartfelt. It feels like you are speaking directly to me. Wow!”

Then Rachel wrote something that stopped me cold: “Reading the chatbot responses to this question, I am perplexed by how much of the language used by the machine is language I would probably use as well.” Her recognition that both humans and AI now operate within a culture that recycles and reproduces ideas so relentlessly that their original meaning begins to fade struck me as unusually self-aware. She seemed to sense that in trying to explain what we don’t fully understand, we risk emptying those very things of their power. In her case, I underlined the passage and jotted, “Brilliant. Astounding. Who are you?!”

Catherine explored this tension from yet another angle: “AI writing is, by definition, nothing new; it’s an amalgamation of pre-existing writing. Of course, one could argue that human writing is also unoriginal and influenced by everything a writer has read, but at least it’s a human doing the work of reading and processing and transforming it into writing.” Though I found her argument compelling, I couldn’t resist gently pushing back in the margins: “I want to agree with you — to believe human moral judgment is something beyond a machine’s reach. But then I think about what people have been doing to each other for… well… forever, and I wonder if we humans are really so adept at moral decisions either. What do you think?”

Simon offered an analogy that deepened the metaphor of reproduction: “Artificial Intelligence is akin to the screen actors who display themselves without any notion of how it will affect those who view it. Humans on the other hand are equivalent to the stage actors who express themselves fully, using raw emotions to appeal to audience members.” This idea — that aura is performative, that it depends on proximity and risk — illuminated Benjamin’s argument in a strikingly original way.

Vera was more skeptical of the aura entirely: “A placebo of meaning,” she called it. And yet even she admitted, “I have felt the aura. I’ve locked eyes with a painting, and felt something tighten in my chest.” What I admired most in her paper was this tension — the refusal to resolve the contradiction. “Maybe the only reason my writing has an aura,” she wrote, “is because you believe that it does.”

The Novel That Kept Appearing in My Margins

A pattern emerged in my feedback: I kept recommending Kazuo Ishiguro’s novel Klara and the Sun to students who expressed too much certainty about the boundary between human and artificial consciousness. This novel, which centers on an artificial “friend” with remarkable emotional intelligence, challenges simple distinctions between human and machine in ways that nudged students toward more nuanced thinking.

When Rebecca confidently asserted that “Everything human is inherently personal, shaped by experience, sensations, and warmth,” I responded in my comments: “This is a strong assertion, but it lacks evidence. Are you sure that will always be the case? Kazuo Ishiguro brilliantly explores this question in Klara and the Sun.” Similarly, to Xavier — whose thoughtful reflection on AI’s limitations I appreciated — I wrote, “Your paper is fantastic. Your voice is powerful, and your style is captivating. Still, keeping an open mind in the face of novel technologies is not necessarily a bad thing. I recommend you read Klara and the Sun because Ishiguro is onto something about what it means to be… well… human — something that may have little to do with having blood, guts, and organic matter.”

When History Becomes a Foreign Country

Many students perceptively identified how the AI responses lacked historical context when evaluating 18th-century medical practices. Katherine pointed out the anachronisms permeating the AI’s answers — how they judged past practices “through modern eyes, in an age with different knowledge, standards, and cultural expectations of what constitutes effective intervention.” Josephine, for her part, captured the essence of historical empathy by noting, “Without [historical] closeness, the AI presents the two cases with no empathy for those involved.” She then quotes Benjamin: “to pry an object from its shell… is to destroy its aura.” In other words, ripping actions out of their historical context fundamentally alters our understanding of them.

Simon echoed this vividly: “There was no connection to them as human beings that existed in time and space,” he wrote. Tessa made the same point analytically: “AI perspective… misses a unique historical argument and instead recycles facts generated from internet sources.” Both noted that AI tends to present conclusions as inevitable rather than contingent, robbing the past of its texture.

Vera argued that Rosenberg’s framework helps restore that texture. Medical care “worked” not because it healed but because it “fit the logic of its time.” Tessa extended this with a memorable formulation: “Physicians relied on provoking visible physical effects in order to demonstrate an effectiveness… Bleeding, vomiting, blistering — all were attempts to restore equilibrium.” The logic wasn’t biological; it was theatrical, ritualistic — a performance of competence that required the patient’s suffering.

The Human Element: Certainty vs. Questioning

Jacob admitted his own bias against AI even as he acknowledged its capabilities: “ChatGPT’s capacity to hoover up knowledge exceeds that of any person — past or present.” In his paper’s margin I wrote a challenge: “What makes you think ‘brain work’ is exclusively your turf? Might a mapmaker have felt the same about GPS, or a tollbooth operator about EZ-Pass? What makes our mental work so special?” Ophelia’s analysis, by contrast, took an evidence-based approach that delighted me. She timed how fast each platform responded: “ChatGPT answered in 22.7 seconds, Gemini in 20.5, and Claude in just 7.8 seconds.” I underlined her sentence and noted, “I love this — one of my favorite lines out of 40+ papers. I love evidence!” However, when Ophelia concluded that AI lacks moral capacity, I found myself hesitating again. “I want to agree,” I wrote, “but remember, people can be terrible too. I’m not sure bots can be ‘immoral’ in the human sense, but I’m also not convinced humans have the moral high ground. This isn’t a question with an easy answer.”

Lucas confessed he felt vaguely ashamed just opening the chatbot sites to do the assignment: “I sat on my bed with my back to the wall so my roommates could not see what I was doing.” His discomfort with the technology itself was telling. When Lucas concluded that AI responses “cannot have an aura as Benjamin described it,” I gently noted, “Perhaps you’re too confident in that claim — maybe taking Benjamin’s idea of aura a bit too literally. I suspect this question will remain open for a long time, if not forever.”

The Wisdom in Complexity

The students who impressed me most were the ones who embraced nuance over certainty. Catherine, for instance, approached the AI prompt with admirable balance: “Could I have produced a better response than the AI? The answer is both yes and no.” And Jonathan acknowledged sometimes an algorithm had contributed to his thinking: “I do not believe I could’ve come up with a better response than ChatGPT because I wouldn’t have appreciated that the flawed methods of late-18th-century physicians were emblematic of the era in which they lived.”

The Space Between Certainty and Doubt

What I value most in these papers is the willingness to engage thoughtfully with complex ideas — to question assumptions, connect disparate concepts to personal experience, and embrace the messy ambiguities of lived reality. While AI can generate factually accurate content with impeccable structure, these students demonstrated that human writing still possesses qualities beyond the current reach of machines. Yet I found my frequent references to Klara and the Sun revealing of my own suspicions: perhaps the boundary between human and artificial consciousness is more permeable — and more philosophically troubling — than many of us want to admit. Instead of viewing AI and humanity as fundamentally opposed, I encouraged students to consider that the very qualities we prize in human expression — empathy, creativity, meaning-making — might one day be mirrored in artificial forms as well. If that day comes, we’ll be forced to reconsider what it truly means to be human in a world teeming with intelligent machines.

The best of these student papers left me not with definitive answers but with better questions. And isn’t that, after all, the hallmark of an education worth having?

Eric Caplan, PhD — Wesleyan University — April 2025