If you're applying to college this year, you're probably wondering: are colleges actually using AI to read my essays?

The short answer is yes—some are. But the reality is more nuanced than you might expect.

We analyzed official statements and documented cases from US universities to find out who's using AI, how they're using it, and what it means for your application.

Quick Answers for Students and Parents

Which colleges confirmed they use AI to read essays?

• UNC-Chapel Hill (since 2019)
• Virginia Tech (starting 2025-26)

Which colleges use AI detection to check for cheating?

• BYU explicitly runs detection software
• UC system runs plagiarism checks, can disqualify AI-generated content
• UW-Madison confirmed they do NOT run AI detection

How common is AI in admissions overall? About 50% of admissions offices use some form of AI, though mostly for transcripts and recommendation letters—not necessarily essays.

Who's Actually Using AI to Read Essays

UNC-Chapel Hill: The Early Adopter

UNC has been using automated essay scoring since 2019, making them one of the first major universities to publicly confirm AI involvement in reading admissions essays.

Here's how it works:

• Tool: Project Essay Grade (PEG) engine by Measurement Incorporated
• Scale: Essays scored 1-4 for writing mechanics
• Purpose: Evaluate grammar and complexity, not content
• Human oversight: Reviewers can override AI scores

According to The Daily Tar Heel's investigation, UNC spent nearly $200,000 on this technology. Rachelle Feldman, associate vice provost, explained the goal: "allow our evaluators to concentrate on the things we think are the most important."

UNC also uses Slate's Reader AI to summarize documents, though they emphasize humans make all final decisions.

Virginia Tech: The New Approach

Starting in 2025-26, Virginia Tech is implementing a hybrid model that pairs human and AI reviewers for each essay.

Their system works differently:

• Scale: 12-point scoring system
• Process: One human + one AI score per essay
• Safeguard: If scores differ by more than 2 points, a second human reviews
• Emphasis: "AI is being utilized to confirm the human reader essay scores, not make any admissions decisions"

Virginia Tech's announcement stressed that the AI model was "trained, rigorously tested for accuracy and fairness" by their own researchers.

The Detection Side: Who's Checking for AI-Written Essays

BYU: Clear Consequences

Brigham Young University has perhaps the most explicit policy. From their admissions page:

• "We use software tools to analyze the admission essays"
• "We may rescind the admission offer" if essays are "found to have been generated by AI"

UC System: Mixed Messages

The University of California presents an interesting case:

• They state they do NOT use AI to review applications
• But they DO run plagiarism checks
• AI-generated Personal Insight Questions can lead to disqualification

Robert Penman from UC admissions was direct: AI-generated responses won't capture what makes you unique, and using them could get you "disqualified from UC admission entirely."

UW-Madison: The Transparent Approach

Wisconsin takes a notably different stance. They explicitly state on their admissions site:

• "We will not disqualify an applicant found to have used or suspected of using AI"
• They're not "running essays through any system to detect AI"

They still discourage using AI to generate essays, but they're transparent about not penalizing students.

Duke's Response: No More Numbers

Duke University made headlines by eliminating numerical essay ratings in 2023-24, citing the rise of generative AI as a key factor. While they still read essays holistically, they no longer assign specific scores—a direct response to AI's impact on admissions.

What About Everyone Else?

Many universities post guidelines about student AI use without revealing whether they use AI detection or scoring themselves. Schools like Brown, Yale, Caltech, Cornell, and Georgia Tech have clear policies about what students can and cannot do with AI, but remain quiet about their own AI use.

According to Inside Higher Ed's reporting, about 50% of admissions offices use some AI in their review process—though this includes transcript analysis and other non-essay applications.

The Technology Behind the Scenes

Most colleges using application management software have access to AI capabilities, whether they use them or not. Slate, the dominant platform in college admissions, offers Reader AI that can:

• Summarize essays and letters
• Highlight key topics
• Pre-score written materials

Just because the technology exists doesn't mean schools use it for undergraduate essays. But the capability is there.

What This Means for Your Application

If You're Applying to UNC or Virginia Tech

Your essays will be scored by AI for writing mechanics. Focus on:

• Clear, grammatically correct writing
• Avoiding overly complex sentences that might confuse scoring algorithms
• Remembering that humans still evaluate content and fit

If You're Applying to Schools with Detection

• Write your own essays—the risks aren't worth it
• Grammar tools like Grammarly are generally fine
• Getting feedback from others is acceptable, but the words must be yours

For All Applications

• Assume your essays might be checked, even if not publicly confirmed
• Include specific, personal details that only you could write
• Focus on authentic storytelling over perfect prose

Why So Few Schools Admit to Using AI: The Transparency Problem

Here's what's fascinating: thousands of colleges have access to AI tools through Slate, yet only two major universities openly confirm using AI to score essays. Why the silence?

The Legal Liability Concern

Schools face a paradox. If they admit to using AI and a student is wrongly rejected due to a false positive, they could face discrimination lawsuits—especially given documented bias against ESL writers. But if they don't disclose AI use, they risk transparency complaints. Most choose silence.

The Competitive Disadvantage

Virginia Tech's announcement is revealing. They emphasized AI would "confirm" human scores, not replace them. Why? Because being the first to say "AI reads your essays" could deter applicants. UNC learned this—they used AI for five years before it became public through journalism, not voluntary disclosure.

The Vendor Lock-In

Slate's Reader AI is available to most admissions offices, creating an interesting dynamic. Schools that don't use it might feel pressure to keep up with those that do. But admitting to using it first puts you in the spotlight. It's a classic prisoner's dilemma.

The Reality Behind the "50% Use AI" Statistic

That widely-cited figure deserves scrutiny. When Inside Higher Ed reports 50% of schools use "some AI," they're lumping together:

• Transcript OCR (optical character recognition)
• GPA calculations
• Recommendation letter formatting
• Essay reading and scoring

The first three are uncontroversial automation. The last one—what parents actually care about—remains rare. Our research found only two confirmed cases of AI essay scoring at major universities. The gap between "has AI capability" and "uses AI for essays" is massive.

What This Actually Reveals About College Admissions

The Efficiency Crisis

UNC's spending $200,000 on AI reveals the real issue: volume. With 70,000+ applications and multiple essays each, human readers face an impossible task. Schools aren't adopting AI because they want to—they're adopting it because they have to.

Virginia Tech made this explicit: their goal is to "review essays more quickly." Not better. Not more fairly. Faster.

The False Positive Dilemma

The most insightful part of this story isn't who uses AI—it's who stopped. Duke eliminating numerical essay scores and Johns Hopkins disabling detection tools shows schools wrestling with a fundamental question: is imperfect AI worse than overwhelmed humans?

MIT researchers warn that AI detectors produce false positives, especially for:

• ESL writers (2-3x higher rates)
• Students with formal writing styles
• Technical or scientific writing

Recent research found concerning bias patterns, with international students and certain racial/ethnic groups more likely to be falsely flagged. This isn't just about accuracy—it's about equity.

The Trust Paradox

UW-Madison's approach is perhaps most telling. By explicitly stating they don't run AI detection, they're making a calculated bet: honesty builds more trust than perfect detection. They're essentially saying, "We'd rather risk some AI-written essays than falsely accuse honest students."

Compare this to BYU's hardline stance—"we may rescind admission"—and you see two philosophies colliding. One prioritizes trust, the other control. Neither has proven superior.

Looking Ahead

The landscape is changing rapidly. In 2019, UNC was an outlier. By 2025, AI involvement in admissions is becoming normalized, though transparency remains limited.

What we're likely to see:

• More schools adopting hybrid human-AI review systems
• Increased pressure for transparency about AI use
• Better detection technology with fewer false positives
• Clearer policies distinguishing acceptable from unacceptable AI use

Key Takeaways

1. Confirmed AI essay readers are still rare (UNC, Virginia Tech), but capabilities are widespread
2. Detection is more common than scoring, with schools like BYU and UC actively checking
3. Transparency varies wildly—some schools are open (UW-Madison), others say nothing
4. The technology exists in most admissions offices, even if unused
5. Writing authentically remains your best strategy

Remember: colleges want to hear your voice, not an AI's. Whether they're using AI to read or detect, the goal remains the same—understanding who you are and what you'll bring to their campus.

---

At GradPilot, we help students navigate the changing landscape of college admissions while maintaining authenticity. Our partnership with Pangram Labs ensures our AI assistance enhances your voice without replacing it.