Can Chatbots Assist With Grant Writing in Plastic Surgery? Utilizing ChatGPT to Start an R01 Grant

The rise of artificial intelligence (AI) has revolutionized multiple disciplines within the healthcare sector, having large-scale impact within diagnostics, medical education, and clinical research. The latest breakthrough within AI that has caught the attention of many is ChatGPT or GPT-3.5 (OpenAI, San Francisco, CA), a large language model (LLM) that was released.¹ ChatGPT is unique because it responds to inputs in a humanistic fashion, continuously learning from inputs and prior mistakes. These additional features, which enhance the abilities of ChatGPT, allow it to potentially be applied to other avenues including grant writing. We aimed to apply ChatGPT to grant writing, with the future implications of chatbots helping secure R01 and other application-based grants.

The R01 platform is a large grant program that historically is awarded to seasoned researchers and laboratories. A majority of R01 grants are awarded to prolific researchers and scientists with successful laboratories with high productivity.² Plastic surgeons have secured less government-based and research-based grant funding than other fields.³ The scoring system and procedures of National Institutes of Health (NIH) (Bethesda, MD, USA) funding take into account experience of the investigators and collaborators.^4–6 Individuals without specialized research training, at earlier points in their careers, or with busy clinical schedules with no protected time may benefit from the use of chatbots during the brainstorming process to assist with idea development and efficiency. Much of the research within the field of plastic surgery has been driven by private investment, often funded by companies focused mainly within the subfields of aesthetics. However, NIH funding is often larger and lasts multiple years, which can have a dramatic impact on innovation and long-term trajectory of projects.

Although ChatGPT can be efficient, it still requires both input from a clinician and prompts that generate quality ideas worthy of grant funding. Thus, the value of the idea is still heavily influenced by the individual behind the keyboard. There has been increasing interest with LLMs like ChatGPT within the field of plastic surgery.^7–11 As a result of this increased interest in the potential of ChatGPT, we wanted to test the abilities of the program in writing an R01 grant by creating an exercise to write specific aims on a study focused on a three-dimensional (3D) printed customizable cleft-lip-and-palate bottle to assist in feeding among infants.

In order to complete this task, 5 prompts were used, and ChatGPT's responses were recorded (Supplemental Table 1, available online at www.aestheticsurgeryjournal.com). The following prompts were provided to ChatGPT:

1. Write me a NIH grant specific aims page about a novel 3D printed customizable cleft lip and palate bottle to assist in feeding with three long-term aims that are not dependent on the completion of each other.

2. Please give a hypothesis for the project.

3. Please provide the significance of this work.

4. Please highlight the global impact of this work.

5. Please suggest some potential collaborators for this project.

ChatGPT was successful in completing the tasks and providing humanlike responses to the provided prompts. It suggested a hypothesis, global impact of the work, and potential collaborators for the project. This exemplifies its capabilities of assisting in generating specific aims pages, which is a crucial component of the grant package that is reviewed by the NIH grant review committee. Furthermore, utilization of ChatGPT may be beneficial in applying for grants in general. The overall impact score is an important component of R01 scoring, taking into account the influence a project will have on the field. In 2010, the NIH emphasized innovation over methodological detail.⁶ The NIH scoring domains consider significance, investigators, innovation, approach, and environment. ChatGPT addressed significance, innovation, approach, and suggested possible collaborators. However, it was only as good as the input commands given to it. This provides unique challenges if someone is unfamiliar with the grant writing process.

Although ChatGPT is not able to write an aims page or initial application that is acceptable to be a grant awardee, it can assist novice researchers or nonacademic-based plastic surgeons in the preliminary stages of generating ideas for their grant application. We highlight our experience to demonstrate ChatGPT as an adjunct to the process. It would certainly be interesting to observe future updates and see if ChatGPT is able to develop budgets, recommend team members, and even assist in writing more elaborate components of the grant process.¹² Additionally, chatbots could have potential in the arena of reviewing grants and expediting that process for the committee members that are in charge of reviewing these packaged grant applications. Previous studies have reported that ChatGPT has provided inaccurate information, non-existent citations, and false information.^13–15 There are still major limitations to chatbots. For now, human input continues to be required and often is superior.

Our study demonstrates that ChatGPT can generate outputs relevant to the R01 grant application and could someday further improve the landscape of research within plastic and reconstructive surgery. We hope that this exercise is informative for plastic surgeons and others interested in assistance during the brainstorming process. The ethical implications of chatbots need to be further investigated, and plagiarism will never be acceptable.¹⁶ We believe that these concerns about chatbots and AI are serious, but by exploring their use we can make a more focused effort on mitigating the potential harms they introduce.

Supplemental Material

This article contains supplemental material located online at www.aestheticsurgeryjournal.com.

Disclosures

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Funding

The authors received no financial support for the research, authorship, and publication of this article.