Building Mental Muscles in the Age of AI

In a recent post, “The Prose and Cons of AI Writing,” I told the story of acceding to a friend’s mischievous experiment. He asked one of the robot writing tools to compose an essay, in my style, about the naughty adult game Cards Against Humanity, keeping my preoccupation with play in mind. (An editor once told me, not entirely uncritically, that I “didn’t so much write sentences as I punch and pound at them.” So maybe I do have a “style.”)

Would the machine simulate believably and persuasively? Could it be witty and direct? Would the device find insights I’d missed? Could it craft an aphorism? “John Henry Steel Drivin’ Man,” the Luddite ballad about the legendary folk hero, rang in my ears as I accepted the robot challenge. I wanted to know: Should I shelve the keyboard and leave writing about play to the machine? Should I lay my hammer down?

This statue of the legendary Luddite folk hero, John Henry, who gave his life in a contest against mechanized railroad tunnel construction, stands near Talcott, WV.

Source: Courtesy Ken Thomas (2001)

To cut a short story shorter, to my ear the AI writing proved pretentious, ponderous, and pompous, and it missed profundity by a mile. I felt not so much parodied as conned. The approximation gave us a good laugh.

Writer vs. Machine

As it happens, this wasn’t the only such vindication. Soon after, “sincere curiosity” prompted the popular novelist Curtis Sittenfeld to accept a similar trial-by-machine from an editor of the New York Times Opinion pages. Could the chatbot manufacture a “lusty, summer beach-read” in Ms. Sittenfeld’s manner? For this experiment, the two asked if readers could tell the difference between machine writing and the writer’s writing by comparing two texts, authorship left unattributed.

The contest began by supplying the machine with elements that included “lust, kissing, flip-flops, regret, and middle age.” Promising starts. No? In just 17 seconds, the robot manufactured a 1,000-word story. The writer took longer to write hers. It took me less than 17 seconds to discern real from robot.

Distinguishing Art From Artiface

The first story, Sittenfeld’s it turns out, starts grabbily: “You can probably see where this is going,” she began. Hmmm. No, we probably can’t. Nor do we know what the “this” is. We readers don’t know where or what. But we wish we did. With that first sentence’s little trick, she has us.

Compare the robot’s opening: “The flip-flop moment. Lydia had always been practical. It was her hallmark…” Discerning readers will note the initial non sequitur and the confusingly weak pronoun reference. Evasive passive voice follows. (Three big mistakes in 12 words, yikes.) The worst part, a fourth fumble: The beginning lacks an invitation. Who cares what comes next? I mean. Why keep reading?

Writer Mops the Floor With Machine

We might stop here with a simple headline: “Writer Mops the Floor With Machine.” But there is more at stake intellectually and morally. With so much information available now and with so little room to maneuver toward originality, recourse to the machine entices fledgling writers. The technology can deliver blindingly fast prose about myriad hot term-paper topics—think cross-border sex trafficking, wars of religion, climate catastrophe, and election subversion. You name it. With this temptation and deadlines looming, drawing the border between assistance and plagiarism proves troublesome.

Professors Freak Out

Lisa Graumlich, a pioneering scientist who charted the effects of global warming and a dean emerita of Environmental Sciences at the University of Washington, wrote me with a capsule history of the way the moral and pedagogical issue of AI impacted the college classroom.

AI “freaked” administrators and professors at first, Graumlich concluded. When academic policies issued dire warnings about deception and unreliable citations, the students “yawned and rolled their eyes.” Teachers freaked some more, predictably. But not so predictably, she writes, “we went from FREAKED OUT to creatively thinking about the future.”

The university formulated a new institution-wide determination to more carefully and personally encourage and nurture how students wrestled with “canonical texts.” The point of education is to grapple with new ideas, to follow facts where they lead, and to coherently synthesize the new knowledge. If AI could help professors pursue these delicate and demanding tasks, help them avoid posing “routine compare-and-contrast questions,” and more carefully integrate lecture, class discussion, and assigned texts,” the new technology would force “faculty to interact with students at a deeper level.”

Artificial Intelligence Essential Reads

We Are All AI's Teachers—And It's Our Job to Teach Wisely

The Future of Instant Communication Through AI

The Deeper Level

Carolyn Korsmeyer, a novelist who recently retired as chair of the Philosophy Department at the University of Buffalo, follows this basic, “absolutely crucial” deeper question, addressing the “changes in how we use our own minds” that AI writing provokes. The world will be better off as AI sorts giant data sets and manages information to deliver stunning advances in fields as diverse as molecular biology and traffic control. But, Korsmeyer notes that learning is more than collecting information. She writes, “if learning includes developing and refining one’s own cognitive and expressive capacities, [AI] can’t possibly be a real learning tool.”

“The effort of crafting ideas into words is of “utmost importance,” Korsmeyer observes, because composition is a process of both recognizing meaning and content and generating meaning and content. Setting down thoughts and strengthening them gives writers a workout. Writers build mental muscle as they write to understand.

Striving to Find the Right Word

To shortcut the route toward discovery is like painting by numbers. I’ll take nothing away from the wry exercise in dexterity and care. Also, it’s useful for decorating wastebaskets. But painting by numbers can lay no claim to originality. The century-old original advertisements for paint-by-numbers sets promised “A beautiful painting the first time you try!" These hobbyists skip struggling to climb the several steps that build insight and visual understanding. They skip art. Real painters explore the relationships between form, color, shading, gesture, material, and subject much the way writers strive to find what we mean and what we think by choosing the right word.

This search is more than a pleasurable word game because discovering what we mean to say and saying it with nuance and consequence resonates with moral undertones in this era of bot-generated falsehood.

The French novelist Gustave Flaubert put it this way: Writing, he observed, is a search for “les mots justes,” the exact words, the right words. Or should we say pointedly now, the just words.