Preservation of Human Creativity in Plastic Surgery Research on ChatGPT
Bibliographic record
Abstract
We read with great interest the latest series by Gupta et al published in Aesthetic Surgery Journal, examining and reporting on promising yet concerning applications of ChatGPT (OpenAI, San Francisco, CA) in scholarly practice.1‐4 ChatGPT is a novel large language and artificial intelligence (AI) model recently released by OpenAI, which has proven capable of interpreting, synthesizing, and outputting information in the form of traditional human text.1‐7 AI represents a rapidly evolving technology within the field of computer science, providing computer systems with the ability to emulate human intelligence and perform human-like tasks.2 Taken together, AI and ChatGPT bring infinite potential through a confluence of capabilities in autonomous perception, knowledge synthesis, as well as inference of information. Unsurprisingly, plastic surgeons have been drawn to the potential for innovation that AI and ChatGPT can bestow upon our specialty. Indeed, publications on the potential role of ChatGPT in plastic surgery have been on the rise. As of May 15, 2023, 19 articles have been published investigating or postulating applications of ChatGPT in plastic surgery. Eight studies (42%) appear to be published by the same group of authors: Gupta et al. and Najafali et al.1‐7 Novelty in research lies not in new permutations of the same idea, but in novel propositions and methodologies with the potential to improve and advance practice, independently of what has been previously published. In a letter published in February 2023, Gupta et al reported on ChatGPT's ability to generate novel systematic review ideas in the field of aesthetic surgery, with variable and imperfect performance demonstrated according to specific topics examined (general aesthetic surgery vs rhinoplasty vs blepharoplasty).1 In March 2023, Gupta et al published again on ChatGPT's ability to generate novel systematic review ideas; this time, for 6 surgical and 6 nonsurgical procedures in aesthetic surgery, again, with variable and imperfect performance reported.2 In the same month, they published in a different journal on ChatGPT's ability to again generate systematic review ideas, now with relevance to different subspecialties in plastic surgery, including cosmetic surgery, craniofacial surgery, microsurgery, and hand surgery.3 They once again reported variable and imperfect performance. Now, and most recently, Gupta et al have published again on ChatGPT's ability, in its new version ChatGPT-4, to generate systematic review ideas, for the same topics previously examined1,4 Again, they report variable and imperfect performance.4 In parallel, Najafali et al published a letter in March 2023, in response to Gupta et al, accentuating the significant ethical limitations associated with the use of ChatGPT in research and scholarly practice, urging “caution when using ChatGPT.”5 Nonetheless, in April 2023, Najafali et al published on ChatGPT's ability to write an entire systematic review on vaginoplasty,6 and later that month, on ChatGPT's ability to write grant applications.7 When analyzing the literature, we must question the value new publications bring to our specialty. The ability of ChatGPT to generate novel systematic review ideas with reference to topics it is provided with has clearly been established, as has its ability to engage in (potentially unethical) scholarly activities, which Najafali et al caution against,5 but also publish on.6,7 Research into the applications of ChatGPT may be stratified and approached with reference to the target emulated human behavior, and the person ChatGPT may be of assistance to in these demonstrated applications. Examples from the former classification include creative thinking, critical thinking, data curation, data analysis, writing, or communication, to name a few. Examples of categories within the latter group include applications that assist the plastic surgeon in her capacity as a researcher/scholar, the plastic surgeon in her capacity as a clinician, the plastic surgery patient, or the plastic surgeon educator and/or trainee. The 6 aforementioned studies all investigate and report on the same capabilities of ChatGPT, the same target emulated human behaviors, and the same (controversial and potentially unethical) applications “assisting” the plastic surgeon researcher/scholar. Regardless of permutations and replications, the conclusion remains the same—ChatGPT is capable of generating systematic review ideas with variable and imperfect performance, and can engage in different levels of scientific writing under human direction. So then, we must ask, what benefit does the next publication bring relative to the prior? The overarching goal of research into AI and ChatGPT remains directed towards closing the gap between postulated utility and adoption. As researchers, the potential that AI can bring to our specialty, and to our patients, is what drives us. To inch our specialty closer to the promise and potential of AI, we must work towards widespread adoption. But before we can achieve this, we must design rigorous and methodologically sound studies on applications of ChatGPT across the array of aforementioned categories of applications. Its performance then needs to be validated and objectively assessed with reference to the highest human standards of care and ethics. Only then will we know whether this technology, in its present form, is suitable for adoption, or whether further developments, refinements, and regulations are necessary—guided by our findings. Creativity may already be endangered by the infringement of AI into academic and scholarly activities. Let us not facilitate this process by maintaining the highest standards of human critical and creative thinking, which have and continue to define our specialty. We commend the authors on their demonstrated passion and productivity through their work, and look forward to future studies they will produce in line with the recommendations provided herein. The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article. The authors received no financial support for the research, authorship, and publication of this article.
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 | 0.003 | 0.014 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.002 | 0.005 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.007 | 0.008 |
| Insufficient payload (model declined to judge) | 0.007 | 0.002 |
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".