Affiliation Bias in Peer Review of Abstracts by a Large Language Model

von Wedel, Dario; Schmitt, Rico A.; Thiele, Moritz; Leuner, Raphael; Shay, Denys; Redaelli, Simone; Schaefer, Maximilian S.

doi:10.1001/jama.2023.24641

Cited by 8 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As OpenAI opened up access to the ChatGPT API, it became feasible to ask the same question many more times. In a recent study investigating whether ChatGPT's peer review conclusions are influenced by the reputation of the author's institution, Wedel et al conducted 250 repeated experiments for each question to mitigate the effects of ChatGPT's randomness [16]. However, not all researchers have recognized this aspect.…”

Section: Main Bodymentioning

confidence: 99%

The Evaluation of Generative AI Should Include Repetition to Assess Stability (Preprint)

Zhu,

Mou,

Hong

et al. 2024

Preprint

View full text Add to dashboard Cite

UNSTRUCTURED The increasing interest in the potential applications of generative artificial intelligence (AI) models like ChatGPT in health care has prompted numerous studies to explore its performance in various medical contexts. However, evaluating ChatGPT poses unique challenges due to the inherent randomness in its responses. Unlike traditional AI models, ChatGPT generates different responses for the same input, making it imperative to assess its stability through repetition. This commentary highlights the importance of including repetition in the evaluation of ChatGPT to ensure the reliability of conclusions drawn from its performance. Similar to biological experiments, which often require multiple repetitions for validity, we argue that assessing generative AI models like ChatGPT demands a similar approach. Failure to acknowledge the impact of repetition can lead to biased conclusions and undermine the credibility of research findings. We urge researchers to incorporate appropriate repetition in their studies from the outset and transparently report their methods to enhance the robustness and reproducibility of findings in this rapidly evolving field.

show abstract

Section: Main Bodymentioning

confidence: 99%

The Evaluation of Generative AI Should Include Repetition to Assess Stability (Preprint)

Zhu,

Mou,

Hong

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…As OpenAI made the ChatGPT application programming interface accessible, it became feasible to ask the same question many more times. In a recent study investigating whether ChatGPT’s peer-review conclusions are influenced by the reputation of the author’s institution, von Wedel et al [ 16 ] conducted 250 repeated experiments for each question to mitigate the effects of ChatGPT’s randomness. However, not all researchers have recognized this aspect.…”

mentioning

confidence: 99%

The Evaluation of Generative AI Should Include Repetition to Assess Stability

Zhu,

Mou,

Hong

et al. 2024

JMIR Mhealth Uhealth

View full text Add to dashboard Cite

The increasing interest in the potential applications of generative artificial intelligence (AI) models like ChatGPT in health care has prompted numerous studies to explore its performance in various medical contexts. However, evaluating ChatGPT poses unique challenges due to the inherent randomness in its responses. Unlike traditional AI models, ChatGPT generates different responses for the same input, making it imperative to assess its stability through repetition. This commentary highlights the importance of including repetition in the evaluation of ChatGPT to ensure the reliability of conclusions drawn from its performance. Similar to biological experiments, which often require multiple repetitions for validity, we argue that assessing generative AI models like ChatGPT demands a similar approach. Failure to acknowledge the impact of repetition can lead to biased conclusions and undermine the credibility of research findings. We urge researchers to incorporate appropriate repetition in their studies from the outset and transparently report their methods to enhance the robustness and reproducibility of findings in this rapidly evolving field.

show abstract

Affiliation Bias in Peer Review of Abstracts

Gallo,

Savage,

Chen

2024

JAMA

View full text Add to dashboard Cite

sults for the different outcomes is not easily explained. There were excesses in distant recurrences and second primary malignancies (non-breast cancer) in the nonpregnancy group (eTable 1 in Supplement 2 of the article). There was also an excess of contralateral cancers in the pregnancy group. Patients with oophorectomy were not excluded from the analysis. In our previous studies, we have seen a benefit of oophorectomy on breast cancer survival in BRCA carriers. [2][3][4] However, Lambertini and colleagues did not report the effect of oophorectomy. The clinical relevance of the effect of oophorectomy is pertinent to young BRCA carriers recently diagnosed with breast cancer who may be choosing between pregnancy and salpingo-oophorectomy.

show abstract

Affiliation Bias in Peer Review of Abstracts by a Large Language Model

Abstract: This study assesses affiliation bias in peer review of medical abstracts by a commonly used large language model.

Cited by 8 publications

References 7 publications

The Evaluation of Generative AI Should Include Repetition to Assess Stability (Preprint)

The Evaluation of Generative AI Should Include Repetition to Assess Stability (Preprint)

The Evaluation of Generative AI Should Include Repetition to Assess Stability

Affiliation Bias in Peer Review of Abstracts

Contact Info

Product

Resources

About