2023, Number 2
Use of ChatGPT in scientific manuscripts
Full text
10.35366/111506
Language: English/Spanish [Versi?n en espa?ol]
References: 5
Page: 65-66
PDF size: 91.76 Kb.
GPT is a type of artificial intelligence developed by OpenAI in November 2022 and designed to generate human-like text. It is a language model often used in chatbots, computer programs capable of holding a text or voice conversation with a human.1,2
Chatbots and natural language processing, including ChatGPT, can revolutionize medical publishing by automating some tasks and streamlining the writing process. It has been mentioned that they can help extract information from electronic records, assist in medical literature searches, or be a guide in writing style and format.
Several experts and medical journals reject the use of ChatGPT since it lacks critical thinking and presents redundant and irrational information; its use means that there are no original ideas and the points cannot be argued; and in the case of using it in a scientific manuscript the content will be from the bot, with this comes medical-legal and intellectual property or authorship problems.2
ChatGPT does not have access to PubMed or Cochrane. Therefore, there are gaps in the information it provides, and it needs to be able to elaborate a structured discussion.2 It can be used to review material, to make a constructive brief but not to make an original blueprint. It requires a human intellectual mind and policies that verify the data generated by artificial intelligence systems.
ChatGPT has already been listed as an author in some scientific articles; some editors think there are better decisions than this. The authorship of an article confers credit for the contribution and responsibility for the content of the work, as stated by the Committee on Publication and Ethics (COPE) and the International Committee of Medical Journal Editors (ICMJE); the latter recommends that for someone to be an author, they must have a substantial contribution, design of the work, interpretation of the data, drafting or critical review, in addition to having final approval of the version to be published; according to this, a chatbot should not appear as an author, but only have an acknowledgment in the article.3
Publishing companies want to create policies for using chatbots, but it is still a controversial topic that needs to be clarified. It could be a tool to detect plagiarism; it would care for language and syntax in articles, and statistical verification would be easier. ChatGPT should be able to tell if a human or computer program did the manuscript.4
It should suggest to journals that the author declare that if artificial intelligence were used to complement the manuscript, it would achieve better transparency.5 The World Association of Medical Editors (WAME), has issued recommendations regarding ChatGPT.
Chatbots cannot be authors; authors must be clear and express how they used chatbots. Authors are responsible for the use of chatbots in their manuscripts. Publishers will need tools to detect content generated by artificial intelligence.
If there are no clear rules, medical journals should not authorize using chatbots or artificial intelligence for manuscript preparation, and, above all, everything related to authorship should be clarified.
REFERENCES
AFFILIATIONS
1 Editor in Chief, Cirujano General. orcid.org/0000-0001-5635-5845
CORRESPONDENCE
Abilene Cirenia Escamilla Ortiz. E-mail: escamillaoa@amcg.org.mx