Polygence vs. Publication-focused Research Programs

"Gone were any errors," Jane-Marie Law, a religious studies professor at Cornell, told The New York Times this spring. "Also gone was a sense of freshness and daring. ChatGPT made everything so safe." NYT, "How A.I. Killed Student Writing (and Revived It)," April 30, 2026

She was talking about her students' essays after they started using AI, honor code or not. I keep coming back to that word, daring, because it's the thing I care about most in a young person's intellectual life and the thing a line on an extracurricular list is the worst at showing. It's what happens when a student puts a half-formed idea on the page and risks finding out it doesn't hold up.

I'd ask families to look for that when they compare research programs. A student can spend months on a question, revise a paper, learn to explain a finding, and still have the paper rejected by a journal. Another student can leave a ten-week program with a publication link. Which one grew more as a thinker?

It's tempting to answer by looking at the outcome. A publication is visible and easy to share, and it can mean something real. On its own, though, it says surprisingly little about what the student learned or how much of the thinking was theirs.

What publication-focused programs are built to produce

Publication-focused programs organize the whole experience around a finished paper that ends in a journal submission and, ideally, a publication credit. I understand the appeal. Families spending time and money want to know what their student will have to show for it, and a citation is a concrete answer. For a student who already has a defined question and mostly needs help preparing a manuscript, a program like that can be useful.

Anyone who has studied narrative knows what happens when an author settles on the ending before writing the story, though. The plot bends to reach it. When publication is the measure of success, a student may pick a question because it looks publishable, or set aside a surprising finding because it complicates the paper they planned. Those choices can get a project submitted on time, and they also close off the parts of research where a student might have learned the most.

Some of these programs say the quiet part out loud and present it as a selling point. They are built to produce papers, and whatever thinking happens along the way is a side effect.

A ten-week manuscript deadline invites shortcuts

Research is slow, and most of the slowness can't be scheduled away. Half the reading turns out to be irrelevant, a method fails, a question has to be narrowed twice before anyone could answer it. If a program promises a publishable manuscript in about ten weeks, all of that plus the analysis and the drafting has to fit inside the window, and something gives.

I suspect that, more and more often, what gives is the student's own thinking.

The numbers point that way. Between May and December of 2025, the share of American students who said they regularly used AI for homework went from 48 to 62 percent, according to RAND polling reported in the Times, and a third said they'd used it to draft or revise their writing. In the same polling, two-thirds of students said the technology hurt their critical thinking. Marc Watkins, who directs the AI Institute for Teachers at the University of Mississippi, told the Times that the research essay students used to write on their own time "is, unfortunately, gone." (NYT) Plenty of high school and college teachers have responded by moving writing back into the classroom so they can watch it happen.

A publication-focused research program asks for exactly that take-home paper, on a deadline, with a publication credit waiting at the end. I don't think most of these programs set out to encourage AI use, and many forbid it. But give a student a fixed deliverable, a short clock, and the knowledge that a chatbot can turn out a literature review in seconds, and that's a lot of weight for an honor code to carry.

Anastasia Berg, who teaches philosophy at UC Irvine, made the strongest version of this argument I've read in a Times essay last fall. (NYT Opinion) She points out that Plato worried writing would ruin our memories. He wasn't entirely wrong, and we took the bargain anyway, since writing is how Plato's dialogues survived to be argued about. Her concern with AI is that the uses most people call harmless, like summarizing or outlining, are the ones where a student learns what an argument is and how it's put together. "No aspect of cognitive understanding is perfunctory," she writes.

That sentence hit close to home for me. I majored in comparative literature, and a lot of the work was exactly those supposedly perfunctory tasks. We'd read two texts against each other, summarize what each one claimed, map where they disagreed, and keep at it until something showed up that neither text said on its own. The summarizing was most of the learning, if I'm honest about it. A student who hands that step to a machine may still end up with a manuscript, and a manuscript produced that way can still be accepted somewhere. They'll have missed the part that would have made them a researcher.

How Polygence approaches research

At Polygence, we start with whatever question keeps pulling a student back, then pair them with a mentor who helps turn that interest into something researchable. The mentor suggests methods, points out weak evidence, and asks the harder questions, but the student makes the intellectual decisions.

A lot of our students start with something as broad as mental health and social media. As they read, the question usually turns out to be too big, so they narrow it, pick a method, run into a limitation, and revise what they're claiming. Adrienne Rich wrote about "re-vision" as the act of looking back and seeing with fresh eyes, and that's a fair description of what's going on. Every one of those changes is part of the research, even though a fixed manuscript schedule would count them as delays.

The Core program typically includes 10 one-on-one mentor sessions, usually spread over 3 to 6 months, so students have time for those detours and the mentor relationship lasts long enough to matter. When a program compresses the same work into ten weeks, it's hard to see how the relationship becomes anything more than formulaic, and at worst it's transactional.

One of our mentors, who has sat on thesis defenses at Harvard, uses a simple test of ownership. Can the student explain why they chose their question and methods, how they got from evidence to conclusions, and what they'd change if they did it again? A polished final product matters. So does being able to account for the thinking behind it, and a program that cares about academic integrity should put as much effort into the student as into the paper.

A journal article is one form among many

We support students who want to write papers and pursue publication, and for the right project a journal submission is a great next step. Revising for an outside audience and taking critique from reviewers who have no reason to be kind is good practice.

Plenty of good questions don't belong in a journal, though. Literature taught me early that form carries meaning, and a sonnet and a novel can take up the same grief and end up in very different places. A student investigating a local problem might write a policy proposal or run an impact campaign. Someone working on an engineering challenge might build a prototype or an app, and someone else might make a documentary or a podcast, or bring the work to a science fair. Our Showcasing tracks (Publishing, Multimedia, Presenting, and Competing) exist so the form can follow the work instead of the other way around.

Families sometimes worry that these routes look less serious than a publication. I'd argue the opposite. Building an app that actually works, or presenting findings to people who are allowed to question them, takes as much rigor as a manuscript and sometimes more. It's also more memorable in a stack of applications full of papers written mostly for the link.

What students take with them

The finished project is the part people see, but most of what a student gains happens before it exists: deciding whether a contradicting source means the question has to change, finding another route when an experiment fails or the data isn't there, explaining a messy idea to a mentor until it makes sense out loud. Every time they make a choice they can defend, they get a little more independent. That doesn't mean doing it alone. Researchers lean on feedback all the time. What matters is whether the student weighs the feedback (and whatever tools they use) and stays responsible for the work.

It usually comes through when a student talks about their project. We ask what surprised them, where they got stuck, or why they trusted one source over another, and we listen to how they answer.

Questions to ask any high school research program

When families talk to research programs, we'd encourage them to look past the list of possible outcomes and ask how a student gets there.

• Who works directly with the student, how are they matched, and what kind of feedback do they give?

• Who chooses the question, the methods, the conclusions, and the final format? And could the student explain those decisions to someone else?

• What happens if the question changes, the evidence points somewhere unexpected, or the project needs more time?

• Does the program support more than one kind of outcome, and how does it help a student choose?

• What's the program's approach to AI, and does the timeline leave room for a student to read, draft, and revise on their own, and what does a mentor do when a draft doesn't sound like the student?

• What will the student be able to do on their own afterward that they couldn't do before?

If publication is a big part of the pitch, ask what "publication" actually means there. It could be help with submission, acceptance by a peer-reviewed journal, or placement on a platform the program works with. If the answer comes down to "we keep submitting until it's published somewhere," the natural next question is what the publication even means. In Alice's Adventures in Wonderland, the Dodo sets up a race with no fixed start or finish, lets everyone run in circles for half an hour, and then announces, "Everybody has won, and all must have prizes." Carroll wrote it as a joke, and it still works as one because every reader understands what happens to a prize that everyone gets. 

A publication can be a real achievement, and we celebrate it when the student understands and owns the work behind it. What I'm hoping for, though, is a student who comes out the other side willing to ask a harder question next time and able to say, with some daring, "I did this work, and I can tell you how."

If you teach or counsel high school students and you're watching this shift happen in your own classroom, I'd like to hear what you're seeing.