JASMA

If you teach in higher education right now, you’ve probably heard some version of this hallway whisper:
“ChatGPT is going to kill original thought. The college essay is dead. Why even bother assigning papers?”
It’s a very 2020s remix of an old panic. Once upon a time, calculators were going to destroy math, spellcheck
was going to ruin writing, and Google was going to make libraries obsolete. Now “regurgitative AI” has taken
the starring role.

The phrase regurgitative AI captures a genuine concern: tools like ChatGPT excel at remixing what
already exists. They are astonishingly good parrots. But parrots aren’t philosophers, and autocomplete isn’t
originality. The real question for faculty isn’t “Will ChatGPT end original thought?” but “How do we teach in a
world where regurgitative AI is the default starting point?”

What Do We Mean by “Regurgitative AI”?

Generative AI models like ChatGPT are trained on massive amounts of existing text. They don’t understand ideas in
the human sense; they predict the next likely word based on patterns in their training data. That’s why so much AI
writing feels a bit like an overconfident student who skimmed the readings and is guessing what the professor wants
to hear.

“Regurgitative AI” is a useful label because it reminds us that these systems:

• Work from what already exists. They can’t sample from tomorrow’s discoveries. Their “knowledge”
  ends at the last training cut-off.
• Remix, don’t originate. They combine and rephrase, but they don’t wake up with a strange,
  nagging question that no one has asked before.
• Lack lived experience. They can describe grief, joy, burnout, or curiosity, but they never
  sit in your office hours, stress over rent, or get goosebumps during a lab breakthrough.

In other words, AI can generate fluent, tidy, and sometimes insightful text, but it cannot want anything.
It doesn’t care if a claim is true or useful. It simply produces what statistically fits. Original thought, by
contrast, is messy and value-laden. It involves surprise, discomfort, and risk. No model has skin in that game.

Why ChatGPT Won’t Kill Original Thought

The worry that generative AI will smother originality assumes that original thought is mainly about
producing words. In reality, the hard work of thinking happens long before the cursor moves and long
after the last period.

Original Thought Is More Than Text Generation

When we say a student’s work is “original,” we usually don’t mean they invented an idea no human has ever had.
We mean they:

• Chose a meaningful question and framed it clearly.
• Evaluated sources instead of copying the loudest one.
• Connected theory to lived experience or local context.
• Made justified decisions and stood behind them.

AI can help with some of the surface work brainstorming angles, suggesting outlines, rephrasing clunky
sentences. But it cannot choose a major, navigate family expectations, juggle three jobs, or decide what kind of
professional and citizen a student wants to become. Those are deeply original projects, and they are still
very human.

We’ve Been Here Before (And Survived)

Higher education has weathered disruptive tools before. Word processors did not end composition; they made it
easier to revise. Search engines did not end research; they forced us to teach discernment instead of pure recall.
Learning management systems did not end face-to-face teaching; they shifted where and how that teaching happens.

Generative AI is different in scale, but not in kind. It accelerates certain tasks and exposes how many of our
assignments have been assessable mainly at the level AI now mimics best: polished but shallow prose, skinny on
genuine engagement. That’s uncomfortable, but it’s also clarifying. If a chatbot can ace a task, maybe the task
was never really measuring the thinking we cared about.

AI Has Built-In Blind Spots

Even the most advanced AI tools still have basic limitations that protect original thought from being “replaced”:

• No access to the room. AI doesn’t hear the sighs in your classroom, the side comments after
  a controversial topic, or the energy shift when you change activities.
• No ability to collect new data. It can help students design a survey, but it cannot go out
  and interview people, run a lab experiment, or observe a local community project.
• No ethical agency. AI can generate an argument, but it can’t wrestle with whether that argument
  should be made, or for whose benefit.

These blind spots mean that as long as our courses ask students to interpret lived experience, gather fresh data,
and make value-laden decisions, there is plenty of room left for authentic originality. The danger isn’t that AI
will do the thinking for us; it’s that we’ll quietly let it.

The Real Risk: Outsourcing Thought to the Machine

While AI can’t eradicate original thought, it can tempt students (and faculty) to skip the hard parts. If a tool
can draft a discussion post, summarize an article, and generate a study guide in seconds, why wrestle with the
material at all?

This is where “cognitive offloading” comes in: letting a system do our mental heavy lifting. Used wisely, it frees
bandwidth for deeper work. Used carelessly, it atrophies the very skills we claim to be teaching analysis,
curiosity, and judgment.

In other words, ChatGPT doesn’t kill original thought. Overreliance on ChatGPT might. The problem
is not the existence of regurgitative AI; it’s whether we design learning environments that ask students to think
with these tools instead of merely asking them to paste what the tools spit out.

Turning Regurgitative AI Into a Learning Partner

So how do we move from panic to practice? The answer isn’t banning AI in frustration or surrendering to it in
resignation. It’s designing courses where generative AI is acknowledged, bounded, and leveraged as a scaffold for
deeper learning.

1. Make AI Use Explicit in Your Syllabus

Students are already using AI tools, whether we mention them or not. A clear policy with concrete examples does
more than any detection software to encourage integrity. Consider:

• Identifying which tasks allow AI assistance (for example, brainstorming or outlining) and which tasks
  require fully human work (for example, reflective journals or in-class writing).
• Requiring students to disclose when, where, and how they used AI in an assignment almost like a lab notebook
  for their thinking process.
• Emphasizing that failure to acknowledge AI use is an integrity issue, not the use itself.

This kind of transparency reframes AI from a secret shortcut into a documented tool, similar to citing a secondary
source or using statistical software.

2. Require Reflections on AI Use

One of the most powerful ways to keep AI from flattening thought is to ask students to think about their
own thinking. A short reflection attached to any AI-supported assignment might prompt:

• What did you ask the AI to do, and why?
• What parts of the output were helpful, misleading, or shallow?
• How did you revise or challenge the AI’s suggestions?

These metacognitive questions help students notice both the strengths and the limits of regurgitative AI. Over
time, they build the habit of treating AI as a fallible collaborator rather than an infallible answer machine.

3. Ask Students to Show Their Prompts and Revisions

When assignments permit AI, have students submit:

• The prompts they used.
• Snippets of raw AI output.
• Their annotated revisions and final draft.

This “process transparency” does several things at once. It discourages blind copy-paste behavior, gives you a
window into how students interact with AI, and turns the assignment into a case study in revision. Students learn
that strong writing and thinking emerge through choices, not just generation.

4. Design Assignments That AI Can’t Easily Fake

If a generic prompt like “Explain the causes of the French Revolution” can be answered with a serviceable ChatGPT
paragraph, the issue is the assignment, not just the tool. Consider tweaks such as:

• Localizing the task: “Explain how the causes of the French Revolution echo in a current debate
  on our campus or in our city.”
• Requiring original data: “Interview three people about their understanding of academic
  integrity in the AI era and connect their responses to course readings.”
• Combining modalities: Have students create a concept map, audio reflection, or in-class
  presentation based on their written work.

These designs don’t make AI irrelevant students might still use it to brainstorm questions or refine wording
but they ensure that meaningful human work is always required.

5. Use AI Critically in Class

One of the simplest teaching moves is to bring AI into the classroom and ask students to critique it. For example:

• Generate a sample thesis statement and have students strengthen it.
• Ask AI for an explanation of a key concept, then have students fact-check and annotate it.
• Have students compare AI’s take on a topic with a scholarly article or lived experiences from the class.

Suddenly, AI is not a black-box answer engine; it’s an imperfect text that students must interrogate. That’s
original thinking in action.

Real Classroom Examples: What Faculty Are Already Doing

Faculty around the world are quietly experimenting with ways to coexist with regurgitative AI. A few emerging
patterns:

• AI-generated drafts as “bad examples.” In some writing courses, instructors feed a generic
  prompt into ChatGPT, hand out the result, and ask students to “grade” and revise it. Students quickly see the
  difference between fluent but shallow writing and work grounded in course concepts and personal insight.
• Version histories as integrity tools. Instructors who worry about AI overuse are leaning on
  tools like document revision histories. Watching an essay evolve over days tells you far more about authenticity
  than a yes/no AI detection score ever will.
• Hybrid assessments. Some courses now pair take-home assignments (where AI use is permitted,
  with disclosure) with in-class checkpoints: quick writes, oral exams, or group problem-solving. Students learn
  that understanding must travel with them, not just live in their browser tabs.

These strategies don’t eliminate regurgitative AI, but they do something more interesting: they turn it into a
mirror. Students see where the tool shines, where it stumbles, and where their unique voices matter most.

What Institutions Need to Do Next

Individual faculty can do a lot, but they shouldn’t have to navigate the AI era alone. Institutions can support
original thought in an AI-saturated environment by:

• Creating clear, flexible AI policies. Instead of blanket bans, develop guidelines that set
  shared values (integrity, transparency, equity) while leaving room for disciplinary nuance.
• Investing in faculty development. Workshops, teaching circles, and AI “sandbox” sessions let
  educators experiment with tools before they show up in student work.
• Revisiting assessment culture. As long as high-stakes grades ride on easily automated tasks,
  students will feel pressure to lean on AI in unhelpful ways. Authentic assessment projects, portfolios,
  performances, and community-engaged work is not a luxury; it’s a survival strategy.
• Centering student voice. Students are not just potential cheaters; they are co-designers of
  responsible AI use. Inviting them into policy discussions turns AI from a threat into a shared challenge.

None of this is quick or simple. But it’s exactly the kind of work universities are built for: taking a complex
new reality and responding with thoughtfulness rather than panic.

Experiences from the AI-Infused Classroom (500-Word Extension)

To see how all of this plays out beyond theory, imagine a few snapshots from AI-infused classrooms.

In a first-year writing seminar, Professor Ramirez starts the semester with a slightly mischievous move. She gives
the class a prompt she’s used for years: “Describe a turning point in your educational journey.” Then she projects
a ChatGPT-generated essay responding to that prompt. It’s heartfelt, grammatically clean, and completely generic.
The protagonist wrestles with “self-doubt,” discovers the value of “hard work,” and thanks an inspiring teacher
all in language that could belong to anyone and no one.

“If this showed up in your peer review group,” she asks, “what feedback would you give?” Students quickly notice
the clichés, the lack of sensory detail, and the absence of any real, specific moment. They don’t need a lecture
on regurgitative AI; they’ve just experienced it. When she asks them to draft their own essays, the room is
noticeably quieter, more focused. The bar isn’t “better than nothing.” It’s “better than the bot.”

Across campus, in an engineering ethics course, Dr. Singh takes a different tack. He assigns teams to design a
policy for AI use in a fictional university, including guidelines for assignments, research, and advising. Students
are allowed even encouraged to use AI to help them brainstorm. But they must document every interaction, paste
representative samples of AI output into an appendix, and write a reflection analyzing where the tool was helpful
and where it was misleading.

During presentations, a familiar pattern emerges: teams report that AI was great at listing generic risks and
benefits (“bias,” “efficiency,” “access”), but terrible at making concrete decisions. When asked to choose between
two competing values like speed versus fairness, or innovation versus equity the chatbot hedged or delivered
slogans. Students, on the other hand, had to make a call. Their policies may not be perfect, but they are
unmistakably theirs.

In a social work program, Professor Lee notices that some students are quietly worried that using AI at all is
“cheating,” even when it’s permitted. Others are overconfident and rely heavily on AI to draft case notes. To
address both extremes, she designs a lab where students use ChatGPT to generate initial case summaries, then
compare those summaries with actual client narratives (with identifying details removed).

The differences are striking. AI-generated notes hit the right buzzwords but flatten nuance: complex family
histories, subtle cultural dynamics, and the emotional texture of the client’s voice largely disappear. Students
begin to articulate their own professional standard: AI might help with structure or language, but it cannot replace
careful listening or relational judgment. That realization doesn’t come from a policy document; it comes from
side-by-side experience.

None of these instructors are pretending AI doesn’t exist. They’re doing the opposite: bringing it into the light,
poking at it, and inviting students to do the same. Along the way, something important happens. Students stop
asking, “Is it okay if I use ChatGPT?” and start asking, “How should I use it and when should I not?”

That shift is the real safeguard for original thought. Regurgitative AI will keep getting better at sounding smart.
Our job in higher education is to help students become the kind of thinkers who can tell the difference between
sounding smart and being smart and who know that their own questions, judgments, and experiences are
still the most irreplaceable part of the learning process.

Conclusion: Original Thought Still Belongs to Us

Generative AI may feel like a tidal wave, but it’s also a mirror. It reflects back the parts of our teaching that
were always vulnerable to shortcuts formulaic prompts, low-authenticity assessments, and grading systems that
reward polished surface over deep engagement. If we respond with bans and fear, we’ll miss the opportunity staring
us in the face.

ChatGPT won’t kill original thought because original thought isn’t located in the tools we use. It lives in the
questions we ask, the risks we take, the communities we serve, and the values we defend. “Regurgitative AI” is a
challenge, but it’s also an invitation to refresh the very things higher education does best: nurture curiosity,
sharpen judgment, and help students write and live ideas that no machine can fully predict.