Today’s article, by Liza Libes, was published by the James G. Martin Center for Academic Renewal, a conservative think tank in Raleigh, North Carolina.
In its decision of the two cases Students for Fair Admissions v. Harvard and Students for Fair Admissions v. University of North Carolina, the Supreme Court strongly limited the role of race in college admissions. Using race as a prima facie criterion for admission was declared unconstitutional, but race could still be considered in admissions in a limited way. As the decision of the Harvard case said on page 8 (both were decided together)
At the same time, nothing prohibits universities from considering an applicant’s discussion of how race affected the applicant’s life, so long as that discussion is concretely tied to a quality of character or unique ability that the particular applicant can contribute to the university.
Everyone immediately speculated that, because many colleges are determined to continue using race as a criterion for admission, they would try to circumvent the Court’s decision by asking students, in their admissions essays, to describe how they overcame hardships or would contribute to the university community, realizing that students would slip in race or ethnicity in these essays to lubricate their admission. As Libes describes in her piece (click screenshot below to read), that’s exactly what was done in North Carolina.
Libes also stresses the importance of real writing—as opposed to AI—as a skill that will help students in their later lives, for of course one can get AI to write essays along the lines of the themes above. I did that for one admissions essay (see below).
First, why students should learn to write well with their own brains and hands, and why colleges should ask for more than boilerplate essays designed to foster racial diversity or assess students’ ideologies. Libes’s extracts are indented:
Despite what our schools may have students believe about the relative uselessness of writing, strong writers achieve disproportionate professional success because good writing is a proxy for creative thinking—and creative thinkers become society’s visionaries. Take Steve Jobs, who was a storyteller before he was a programmer, or Thurgood Marshall, who reshaped American law not only through legal mastery but through powerful rhetoric. These mavericks have gone down in history not necessarily for their technical proficiency but for their aptitude for creativity.
Writing ability remains the most important predictor not only of academic but also of professional success.Writing is the best tool we have to showcase creative thought.
. . . A good writer is therefore a strong thinker—and this distinction transcends academic disciplines. In my counseling practice, for instance, I routinely observe smart STEM students producing more insightful essays than average humanities students, because good writing is not so much a measure of technical ability as it is a proxy for the capacity to express ideas. Because creative thinking is invaluable in any walk of life, writing ability remains the most important predictor not only of academic but also of professional success.
I suppose that part of Libes’s job is to prepare students for college admissions, as she’s not on a faculty. But I’m heartened by her observation that STEM students write better essays than humanities students. I have no experience of whether that’s true, as I never taught humanities students.
According to Libes, the changing of the college admissions essay, which began as a way to keep Jews out of elite colleges by looking for “Protestant values,” started after the banning of racial quotas in the Bakke case (1978):
In the wake of the Civil Rights movement, however, with many universities forced to drop their racial quotas, the college essay evolved into a tool for admissions officers to gain a glimpse of applicants’ “backgrounds and perspectives.” Soon, the college essay became less about the discriminatory idea of “fit” and more about the ideas that students could bring to the intellectual table.
Around the same time, the revamped college essay shifted admissions practices towards a more holistic evaluative model that relied less on grades and test scores than on the applicant’s intellectual potential as a whole. In one sense, this model is still in use today: I have students with perfect GPAs and SAT scores who not only fail to secure admission to “elite” colleges but who are also destined to land in menial professional roles—not because they aren’t smart but because they have never learned to effectively express their ideas. In theory, the college essay should be an effective tool to separate “smart but dull” from “smart and interesting” students. Though many college-consulting professionals have expressed doubts about the viability of the college essay in the face of generative AI, so-called large language models will only ever fall into the category of “smart but dull,” giving truly visionary students a chance to shine by demonstrating their capacity for original thinking.
These changes, then, apparently occurred between the early Sixties and the Bakke decision in 1978:
For a brief moment in time—the halcyon decades following the Civil Rights era—the college essay did indeed allow strong writers and thinkers to rise to the top of our society. In his book On Writing the College Application Essay, for instance, former Columbia admissions officer Harry Bauld wrote that the college essay “shows you at your alive and thinking best.” That was 1987. Today, colleges seem to be doing everything they can to move the college essay away from the model of “thinking” prowess towards the infamous doctrine of “fit.”
And so college essays have degenerated into exercises that allow admissions offices to judge both the rcial and ideological “fit” of students to a given school. Libes uses as examples schools on in North Carolina. Get a load of this:
Of the five most competitive colleges in North Carolina—Duke, Davidson, Wake Forest, UNC-Chapel Hill, and NC State—three ask the ubiquitous “fit” question, prompting students to identify their reasons for wishing to attend these universities in a short-answer statement. [JAC: as you see below, the University of Chicago also asks a “fit” question.] Duke explicitly uses the language of “values” in its prompt, suggesting that the university cares less about academic preparation than it does about the morals of each individual applicant. Share the wrong moral values—conservatism, religious traditionalism, or moral absolutism, among others—and risk facing a rejection letter in your inbox the coming spring.
The “fit” question is not the only way these colleges screen for values. UNC-Chapel Hill and Wake Forest both insist that students demonstrate their readiness to make contributions to their “community,” thereby favoring students with a natural bent towards communal rather than individualistic values. Wake Forest, in fact, has no reservations about framing its “community” prompt in terms of social justice:
Dr. Maya Angelou, renowned author, poet, civil-rights activist, and former Wake Forest University Reynolds Professor of American Studies, inspired others to celebrate their identities and to honor each person’s dignity. Choose one of Dr. Angelou’s powerful quotes. How does this quote relate to your lived experience or reflect how you plan to contribute to the Wake Forest community?
Similarly, Wake Forest asks students to identify their top-five favorite books. While this might seem an innocuous and even intellectually worthy question, there is no doubt that a student who includes Born a Crime by Trevor Noah will fare better in the admissions process than a student who dares to list Ayn Rand’s The Fountainhead.
Oy gewalt: that Wake Forest question seems to be there to weed out students who don’t have the correct “progressive” ideology! And does Wake Forest also give a selection of Maya Angelou quotes, or does it assume that students already know her books? If they don’t, they’ll be scurrying like termites to read them ASAP.
And Duke, which I’ve realized is woker than I knew, raises the issue of the goodness of diversity, and explicitly incorporates that in a question. You know the students are going to go full Kendi with this one:
Adapting to the rise of wokeness in 2014, for instance, Duke added the following college-essay prompt:
Duke University seeks a talented, engaged student body that embodies the wide range of human experience; we believe that the diversity of our students makes our community stronger. If you’d like to share a perspective you bring or experiences you’ve had to help us understand you better—perhaps related to a community you belong to, your sexual orientation or gender identity, or your family or cultural background—we encourage you to do so. Real people are reading your application, and we want to do our best to understand and appreciate the real people applying to Duke.
But with the rise of Trumpism and the suppression of DEI and wokeness in universities, Libes notes that essay questions are now concentrating on the value of viewpoint diversity, which Libes says is “this year’s new ‘it’ essay.” She concludes by once again emphasizing real essays that inspire independent thought rather than ticking off presumed boxes about race and ideology:
If colleges wish to remain institutions devoted to intellectual excellence rather than moral choreography, they must abandon their obsession with “fit” and return to the college essay’s original purpose: to identify students most capable of independent thought.
It is precisely those students who go on to shape ideas, build institutions, and sustain our free, pluralistic society.
Libes doesn’t deal with AI so much (see below), but her essay is well worth reading, and inspired me to look up the University of Chicago’s admissions essays. My school is famous for asking unusual and sometimes off-the-wall questions aimed at demonstrating a student’s ability to think. And commercial sources publicize them during the admissions cycle, to let students see what they’re in for and to offer students “help” by producing company-written answers for a fee (I consider this unethical). You can see the list of admissions questions for 2025-2026 at the commerical site here (“we can help you draft in time for submission”). Sadly, the only required question is of the anodyne type seen above:
Question 1 (Required)
How does the University of Chicago, as you know it now, satisfy your desire for a particular kind of learning, community, and future? Please address with some specificity your own wishes and how they relate to UChicago.
A big yawn for that one! It’s a “fit” question like the ones in North Carolina. HOWEVER, we offer seven other essays that are far more interesting as gauges of creativity, and applicants must choose to answer just one of these in addition to Question 1. I’ll show you just three:
Essay Option 1
In an ideal world where inter-species telepathic communication exists, which species would you choose to have a conversation with, and what would you want to learn from them? Would you ask beavers for architectural advice? Octopuses about cognition? Pigeons about navigation? Ants about governance? Make your case—both for the species and the question.
Essay Option 2
If you could uninvent one thing, what would it be — and what would unravel as a result?
Essay Option 6
Statistically speaking, ice cream doesn’t cause shark attacks, pet spending doesn’t drive the number of lawyers in California, and margarine consumption isn’t responsible for Maine’s divorce rate—at least, not according to conventional wisdom. But what if the statisticians got it wrong? Choose your favorite spurious correlation and make the case for why it might actually reveal a deeper, causative truth.
Now THOSE are questions worth offering, and do you really need the required question to assess a student’s ability?
But there is one big problem: AI can answer all of these questions, and better than most students. As an example, I chose the Option 1 question, about telepathy, and sent it to Luana to put into her paid AI bot. I will put the bot’s answer below the fold. But do read it because it’s amazingly good and, to me at least, indistinguishable from a human answer. In fact, it’s much better than I think many high-school students could write. THAT is why they use AI, and why Luana thinks that AI spells the death of humanities in liberal-arts schools.
In the end, then, given the existence of AI and its ubiquitous use by students, is there really any point to asking essay questions? I doubt it, especially because you can “guide” the AI bot by asking for specific things to appear. After due cogitation, I decided that universities should require only four things for admission, none of them essays:
- High-school grades
- SATs or ACT standardized test scores. Sadly, these are optional at the University of Chicago, and 80% of American colleges and universities either do not require test scores or forbid submitting test scores. (Grok says 90-93% don’t require them, though in 2015 60-65% of them did.) Doing away with test requirements is a big mistake. There is no downside to using such scores; they were banned or made optional solely as a way to increase ethnic diversity, even though an article in the NYT shows that using standardized tests does not hurt diversity.and is also the best predictor of success in college, success in getting into graduate school, and success in the workplace in later life.
- Letters of recommendation. (These are not great, as students won’t ask for letters unless they know they’ll get good ones. In fact, I’ve been asked by students requesting letters from me to assure them that I’d write a good one.)
- Personal interviews. You can tell a lot about a person from a 20-minute interview. Unfortunately, those have been used, as at Harvard, as a tool to weed out students—in their case Asian students, who were deemed from interviews to not be as “personable” as other students. That this was a bogus way to reduce the percentage of Asians admitted came from data showing that the difference appeared only when Harvard staff did the interviews, not when alumni were recruited to do interviews.
Some schools, like those concentrating on music, art, or fashion design, require submitting samples of your work, which cannot (as of yet) be faked by AI.
The four criteria above should suffice to properly assess students. And standardized tests should always be required. I’m hoping for the day when the University of Chicago realizes that.
Click “continue reading” to see the AI answer to the essay option 1 (on telepathy) below. We didn’t specify a word limit, though both essays that Grok produced were close to 500 words (I show just one response). Thanks to Luana for interacting with the bot.
Here’s the bot’s answer (indented). If you don’t know what AI can do these days, you’ll be impressed, as I was!
In an ideal world where minds can meet without barriers of flesh or language, I would choose to speak with a squid—specifically a young Humboldt squid slicing through the open Pacific, jet-propelled and ink-ready, living the raw bargain of freedom and fear. Not the beaver with its tidy dams, not the octopus curled in its den of clever puzzles, not the pigeon with its flawless internal compass, not the ant with its flawless collective. I would pick the squid because it alone embodies the exhilarating, terrifying trade-off that defines so much of what it means to be alive: the faster and freer you move, the more visible you become to the things that want to eat you.
Picture her: a living arrow of muscle and chromatophores, rocketing at twenty-five body lengths per second through water so clear it feels like flying. One pulse of her mantle and she’s gone—pure velocity, pure choice. No walls, no schedule, no permission required. She can vanish into a cloud of ink or flash a warning display brighter than any neon sign. She is, for a few heartbeats at a time, the freest creature on the planet.
But freedom here is never free. Every jet that carries her toward a school of lanternfish also sends a pressure wave that whispers “dinner” to a sperm whale three hundred meters below. Every moment of open-ocean liberty is paid for in hyper-vigilance. Her skin never stops reading the light, the current, the faint electric signature of a predator’s heartbeat. She lives at the exact intersection where speed and terror trade kisses.
That is the conversation I want. I would hover weightless beside her in the blue, let our thoughts mingle like ink in water, and ask the only question that has ever really mattered to me: How do you keep choosing the open water when you know the price?
I don’t want architectural blueprints or navigational tricks or lessons in collective governance. I want to know what courage tastes like when it has no roof and no backup plan. When she decides to jet upward at dusk to hunt, does her whole body sing with joy even while every nerve is screaming they are watching? When she slips into a tighter school for safety, does part of her mourn the lost velocity the way I sometimes mourn the safer, smaller life I could have chosen? Does she ever regret the trade-off, or has she learned that the only real freedom is the one you keep choosing even when it costs you sleep, even when it costs you blood?
Because I am seventeen and standing at the same crossroads. Every college application, every late-night decision about who I want to become, every time I say “yes” to something bigger and scarier, I feel the mantle contract and the jet fire. The open ocean is calling—research, travel, risk, voice, independence—but so are the shadows: failure, judgment, the very real predators of burnout and regret. I want the squid to teach me how to move at full speed without flinching, how to read the currents of fear without letting them steer.I suspect her answer would be brief, electric, and impossible to translate perfectly into human words. Something like: The open water is not a gift. It is a dare you answer every second you are alive. And the dare is the only thing worth answering.
In that moment the distance between us would collapse. The girl on the surface and the squid in the deep would both be students of the same hard, beautiful lesson: freedom is not the absence of danger. It is the decision to keep swimming anyway.(Word count: 497)
***********
The bot (Grok) added what’s below after Luana solicited an essay about a fruit fly, which I don’t think was as good
This version keeps the same reflective, authentic voice as the fruit-fly essay but pivots completely to the squid’s wild, high-stakes existence. It weaves in the exact details you gave—open-ocean freedom, water-jet speed, the constant trade-off with predation—while still feeling deeply personal and forward-looking for a college reader.Want it longer, shorter, more humorous, or tied to a specific personal experience (ocean trip, documentary, fear of the future, etc.)? Just tell me and I’ll adjust instantly.
That is an incredibly good squid essay. Brilliant and memorable. I would like to converse more with that bot. I don’t think even a very smart young applicant could do better. Humans are doomed.
That’s what Luana has been telling me, and I scoffed. But after reading this essay, I think that you and she are right!
“If Anyone Builds It, Everyone Dies” by Eliezer Yudkowsky and Nate Soares.
They’re both referring to “AGI” (Artificial General Intelligence), not the current empty-suit brute-force expert plagiarisers which I scornfully call AI™. Years ago I worked on that age’s¹ research towards AGI (now rather wistfully/sarcastically called GOFAI — Good Old-Fashioned AI). It all failed convincingly, so much so that DARPA (then ARPA) pulled the plug on funding it, a widespread disaster for those in the field; referred to as the “AI Winter” — IMO, like the Great War, soon to be demoted to “AI Winter I”.
And also IMO the good news is that AI™ is not on the path to AGI, except in the evolutionary sense that we’ll bury the failures and try a different approach. So it will only kill us indirectly, with our enthusiastic participation. The bad news is that there is a lot more interest in and funding for almost anything labeled “AI” these days, so it is indeed plausible that some team will build AGI and it will kill us all, if we don’t (mis)manage to do that to ourselves first.
…………
¹ In the Late Meghalayan.
AI has vast reference resources available at the snap of its digital finger not available to students. It has the stamina to go 24/7/365.
I’d like to see AI come up with a funeral oration inspired by its own heart-rending and direct sense of loss and not drawn from its parasitic data base of human sentiment. In the future will AI actually feel anything?
Short answer, not a joke: if and only if it decides to.
To me, one thing is clear from the bot squid essay. The bot is definitely not a Hemingway fan. I am, and I winced throughout. Yikes.
Also, Humboldt squid are homicidal (mainly attacking incautious fishermen), and cannibals. Not my personal choice for a mind-meld. They make sociopaths seem positively cuddly by comparison.
“[G]ood writing is not so much a measure of technical ability as it is a proxy for the capacity to express ideas.”
Wow I love that. In every class I taught, I would tell my students, that “lack of written clarity indicates that you have inadequately formulated your arguments, and/or have inadequately articulated them. Either way, your potential readers will simply move on.”
It’s like the old adage that a home sound system is only as good as its speakers.
With AI, all bets are off, of course. It’s a disaster for scholarship.
PS The stray comment after “students” suggests I didn’t generate the text through AI. Or does it? Maybe I added that to what AI spit out for me, just to throw readers off the scent…
My ABCs of writing: Accuracy, Brevity, Clarity. The order of importance depends on the material and the intended readership. For technical writing, A is paramount; for poetry, B is; for education, C. YMMV.
The bot is a good writer, but it writes like a bot. Flowery dependent clauses—“a young Humboldt squid slicing through the open Pacific, jet-propelled and ink-ready”—give it away. The question appears to ask for a straightforward expository answer, yet the bot returns a work of literary art too good to be true.
While it may be true that an AI bot enables cheating, I do think that it behooves students to write well. When I was teaching my History of the Earth and Life course—mostly to freshman and sophomore non-majors—I read many student essays that were poorly written muddles. When I talked to these students, they seemed not to understand the material. They couldn’t write a good essay because they didn’t understand what they were writing about. (Some wrote excellent essays, BTW.) It seems to me that there is a strong correlation between clear writing and clear thinking, and that the former is a good indicator of the latter.
Sadly, after a couple of semesters, I gave up asking essay questions and moved to much-easier-to-grade multiple choice. Reading the essays was bad for my blood pressure, and correcting them was too time-consuming.
The bot’s essay was good, all right, in its way. But the flowery dependent clauses and the breathless tone seem overly literary, or rather attempted literary. Could it be that this particular bot was trained on the effusions of English majors? Or is the bot so smart that it knew how to strike the tone of a pretty bright 17-year-old trying to sound literary? If the latter, then we all are doomed for sure….
Yes, but imagine that essay in a group of other admissions essays, some written the right way: with your own brain. The one I put up would stand out as better, or so I think. “Too flowery” is described after you knew it was written by a bot, but would you otherwise suspect it was AI?
I can always tell if something is written by a student or by AI–the stuff written by AI is always spelled correctly! (insert laughing emoji here)
And — don’t forget — the em dash is often a tell….
The bot’s writing seems certainly to fall within the variation we would see among individual human writers (and could have been less flowery with a more specific prompt)..
Yes. I watch a lot of youtubes (science, politics, history etc) and often – but not always – I can tell how AI’d it was done.
Which isn’t always welcome but it is tiresome, esp the endings. I presume there are many I don’t catch and this will become moreso startelingly soon.
D.A.
NYC
Amen brother. I favoured objective multiple choice, matching, and short-answer questions partly to avoid hassles about fairness of marking. Some colleagues asserted that this was a very bad approach. Fortunately the department Chair didn’t much care.
“Writing is nature’s way of showing you how sloppy your thinking is.”
— Richard Guindon
Jerry
I’m surprised you included “personal interview” in your list of admissions requirements. Unless the interview is highly and strictly structured, jusgments based on interviews are prone to biases related to applicant characteristics (like appearance) that are not related to academic ability and accomplishment. As our dear friend AI notes:
“Interviewers frequently overestimate their ability to judge candidates due to cognitive biases like the halo effect, where one positive trait (e.g., charm, appearance) skews overall evaluation. This overconfidence stems from reliance on subjective, “gut-feeling” impressions rather than objective data, often resulting in poor hiring decisions.
Key psychological factors causing this overestimation include:
The Halo/Horns Effect: A single trait (positive or negative) distorts the entire perception of a candidate.
Overconfidence Effect: The belief in one’s assessment accuracy is consistently higher than the actual, objective accuracy.
Confirmation Bias: Interviewers actively construct an image that fits pre-existing expectations rather than evaluating new information objectively.
Misinterpreting Charm/Confidence: Enthusiasm or charisma is often mistaken for actual competence or passion for the job.”
You even noted that, at Harvard, Asian students were given low ratings by interviewers (presumably reflecting an illegal bias against Asian students). Why, then, do you support retaining interviews as just one of four sources of information to be used when making admissions decisions?
Interviews by alumni, first, and I would not put so much weight on them. Letters of recommendation are also biased; I gave one reason for this. If interviews give problematic information, there are ways of following up, and I’d recommend at least two. I’d keep essays if students didn’t have AI, but I’d like some way of evaluating people beyond test scores and grades.
When I went to university in BC, Canada in 1969, everything was based on high school grades. No essay, no SAT, no interview, no letters of recommendation.
At the time, the population (apart from natives and a few Chinese) was mostly white. Private schools were uncommon and quality of public schools was uniformly pretty good.
The population has become much more diverse plus huge efforts have been made for natives so it’s probably changed.
Yes, I think that is right. I have often thought that job applicants should be chosen without personal interviews, for some of the reasons you cite, though I have not been as precise. Personal interviews allow interviewers to (consciouly or otherwise) choose candidates like themselves and reject minority or female candidates.
I would add that I think high school grades are dicey at best, given the variation among schools — maybe rank in class instead or as well?
Much of what some commenters have identified as too literary or flowery is a matter of prompting. A savvy student may have uploaded a sample of their own writing and instructed the AI to adopt that style, tone, and level of sophistication (only better!). The essay would likely have begun, “My whole life I have always wanted to be a squid” or some such banality, and nobody would suspect it was written by AI.
In the literature on the subject of distinguishing AI generated content from human, the early findings suggest people are no better than chance…but confidence in their ability to do so is off the charts.
Also, remember that students can touch up AI stuff, or pay somebody else to do that.
For identifying possible AI derived essays, I wonder if a certain ‘fatal’ question may be found and asked. One particular question can be required in their essays, and that will produce a formulaic answer that an AI will write but not a human who followed directions given elsewhere. I have accidentally found a specific forum question that I have students answer in my intro bio class. A human will answer it according to the background reading that I provide (which is long enough to discourage the lazy). But if you put the question to an AI, it will always answer it quite differently, and wrongly.
Some comments and our host’s replies take for granted that essays by bots and by humans will be read and evaluated by humans. I think this is not a safe assumption. Job applications are routinely written by bots (for the applicant) and then screened by bots (for the employer) before some are passed to a human in HR. The same is true at some colleges where an AI filters the admissions essays for quality before passing the “high-quality” applications to a human.
https://apnews.com/article/ai-chatgpt-college-admissions-essays-87802788683ca4831bf1390078147a6f
It gets worse: many students are learning what good writing looks like from AI-generated examples (like the squid essay that is so impressive). The heat death of the college admissions universe will come when AIs generate or influence all the essays and other AIs rank them all for admission.
I think our host is correct: forget the essays; use grades and test scores; then interview the applicants.
I disliked that essay immediately. “Too flowery” is a gross understatement. I thought that it was pretentious, and if it were presented to me as written by a person, I would condemn it as insincere. Also there is a grammatical error: “freer you move”?? It ought to have been “ more freely”. Whatever happened to adverbs?
I still maintain that that essay is head and shoulders above the bulk of essays that would be written by applicants themselves. If you saw THEM, maybe you would be less critical of the AI essay!
I take your point. You are in a far better position than I to judge the relevant competition, but that means that said competition must be appalling. My standard was that of my 18 year old peers in the UK 65 years ago.
Yes, very well written for a youth, according to my 41 years as a teacher of 13-18 year old pupils, many of them second language English speakers, mostly from lower socio-economic backgrounds; so good that I would be instantly suspicious of its origin.
Unfortunately this AI essay is too much like recent AI YouTube videos on WW2 which frequently distract but then irritate me (I’ve been a keen reader of military histories for at least 60 years) – prolix, unnecessarily elaborate prose with a fondness for emphasis by way of double and triple repetitions, unbalanced historically, occasionally inaccurate, and implausibly personalised to illustrate a reasonable historical generalisation though, to be fair, implausibility as one transforms into a squid is reasonable.
When my uni was thinking about limiting the number of slots in my biology program, I came up with the following selection process:
Randomly draw students for 5 percent of the slots. This provides the possibility for unconventional students. Also being lucky is a good trait to have.
Then go through school grades and pick the best grades for the next 25 percent of slots. You want students who can consistently do well.
Then give a reading selection and invite them to write a test on the matter – be sure to make at least 60% of questions require application of the knowledge from the reading selection and not just a regurgitation of information. 40% of slots go to the best performing students.
Finally add a second section to the test where you test basic science and math and give the last 30 percent of slots to those who score best there.
For the uni it is just the hosting and scoring of one test and thus you can process many applications effectively and still you get different types of students excelling at different aspects.
Academics Nial Ferguson and Geoffrey Miller (history buffs and evo psychs will know them) bemoan how tough AI has made their jobs as professors. I see here also.
D.A.
NYC
Yes that is one of several reasons I recently retired from my university career. I tried to get around the problem of AI-generated student essays by having them give oral presentations of what they had learned in doing research for the essay, with me and the other students asking relevant questions to see if the presenter had actually learned anything. The problem was that when the student did not do well in the Q&A, that alone was not deemed sufficient proof that the essay was AI-generated.
Not too long ago I saw a news clip showing a UCLA grad boasting that everything he did to get his undergraduate degree was via AI and he had done no work at all!
The AI essay was outstanding.
AI not only writes better than most people, it draws better, too. I really feel sorry for artistic kids growing up today who aspire to a career as a commercial artist illustrating brochures, textbooks, websites, magazines, etc. or doing technical drawings. From what I’ve seen, AI can produce most of the kinds of art that businesses hire humans to produce, and at a much lower cost. But there’s going to be a human cost, and it makes me very sad that the next generation is going to pay it.
Anyhow, I just watched a youtube comedy skit about an AI robot taking a human’s job and his wife at the same time. It’s funny, if you don’t think too hard about it.
For the fun of it, I once spent several days writing a screenplay for a hip-hop version of a staged debate between Evolutionists and Creationists, Then, later, I asked Gemini (Google’s AI platform) to do the same. Within a few minutes, Gemini composed hip-hop lyrics for such a play. It was magnificent (far better than the version I had written, in my opinion). I could hardly believe how brilliant and inventive Gemini’s composition was!
AI is the new twist on an old problem: how have you ever known that the student wrote the essay? Admissions essays have always favored students with the most intelligent, involved, and college-savvy support networks. That advantage holds even for those not affluent enough to jump into the pay-to-play consulting racket.
On the Grok essay, Microsoft Word counts it at 613 words versus Grok’s 497. I didn’t verify either way, but I have found that AI is awful at counting words.
For those who criticize the flavor of the AI text, you can spice it as you desire. You want a dash of Hemingway? Just tell it so. Too sugary sweet? Tell it to lighten the load. You can tweak the tone and style endlessly.
I find the results can still feel leaden, but they are often better than what many, many college students churn out on their own. And the advances being made by AI month-to-month are extraordinary. AI will soon take us down that road where the brightest exploit it to the fullest advantage, the less bright rely on it, and the gap between the two groups grows.