Re-evaluating the role of personal statements in pediatric residency admissions in the era of artificial intelligence: comparing faculty ratings of human and AI-generated statements
Bibliographic record
Abstract
Background: Personal statements play a large role in pediatric residency applications, providing insights into candidates' motivations, experiences, and fit for the program. With large language models (LLMs) such as Chat Generative Pre-trained Transformer (ChatGPT), concerns have arisen regarding how this may influence the authenticity of statements in evaluating candidates. This study investigates the efficacy and perceived authenticity of LLM-generated personal statements compared to human-generated statements in residency applications. Methods: We conducted a blinded study comparing 30 ChatGPT-generated personal statements with 30 human-written statements. Four pediatric faculty raters assessed each statement using a standardized 10-point rubric. We analyzed the data using linear mixed-effects models, a chi-square sensitivity analysis, an evaluation of rater accuracy in identifying statement origin as well as consistency of scores amongst raters using intraclass correlation coefficients (ICC). Results: There was no significant difference in mean scores between AI and human-written statements. Raters could only identify the source of a letter (AI or human) with 59% accuracy. There was considerable disagreement in scores between raters as indicated by negative ICCs. Conclusions: AI-generated statements were rated similarly to human-authored statements and were indistinguishable by reviewers, highlighting the sophistication of these LLM models and the challenge in detecting their use. Furthermore, scores varied substantially between reviewers. As AI becomes increasingly used in application processes, it is imperative to examine its implications in the overall evaluation of applicants.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.074 | 0.290 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.001 | 0.004 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".