Low-Income Students More Likely to Submit AI-Generated Admissions Essays

Key Takeaways

• Lower-income applicants more likely to use AI for essay drafts
• AI use correlates with higher rejection rates among low-income students
• Free-tier AI tools produce lower-quality, more homogenized essays
• Wealthier students access premium AI, counselors, boosting essay effectiveness
• Essay homogenization rises post‑AI, reducing personal narrative diversity

Pulse Analysis

Artificial intelligence has become a staple in college‑admissions pipelines, with recent surveys indicating roughly half of applicants use AI to brainstorm and one‑fifth to draft their personal statements. The Cornell‑Carnegie Mellon analysis adds nuance by linking AI usage to socioeconomic status, revealing that students who receive application fee waivers—an established proxy for low income—turn to free‑tier models like ChatGPT more often than their wealthier peers. This reliance on limited‑capacity tools not only yields lower‑quality prose but also coincides with a higher likelihood of rejection, suggesting that AI is not a neutral equalizer but a new vector of disparity.

The study’s data also highlight a striking linguistic convergence: after the rollout of generative AI platforms, essays grew increasingly homogeneous, especially among low‑income and rejected applicants. Homogenization erodes the personal voice that admissions committees traditionally value, turning the essay from a showcase of unique experience into a templated narrative. For affluent students, access to premium AI subscriptions—such as Claude’s $200‑per‑month tier—and professional essay coaches mitigates this effect, allowing them to harness AI’s strengths while preserving individuality. This bifurcation mirrors the broader digital‑divide framework, where unequal access to technology translates into unequal outcomes.

Policymakers and admissions officers must grapple with these insights to preserve equity. Potential responses include adjusting essay prompts to demand demonstrable personal reflection, incorporating AI‑detection tools calibrated for socioeconomic bias, or offering free premium AI access to under‑served applicants. Further research should dissect the specific linguistic markers that differentiate authentic versus AI‑generated narratives across income groups. By proactively addressing the intersection of AI and educational stratification, institutions can safeguard the essay’s role as a genuine window into an applicant’s character and experience.