Many imaging journals still lack policies on the use of AI in manuscript writing
Despite concerns over the factuality of artificial intelligence-generated work, a significant portion of medical imaging journals still do not have policies regarding the use of AI when submitting manuscripts.
In fact, nearly 40% of MEDLINE-indexed radiology journals do not provide guidelines for AI generated content, according to a new analysis published in Academic Radiology.
“Artificial intelligence technologies are rapidly evolving and offering new advances almost on a day-by-day basis, including various tools for manuscript generation and modification,” corresponding author Onur Simsek, MD, with division of neuroradiology at the Children’s Hospital of Philadelphia, and colleagues noted. “On the other hand, these potentially time- and effort-saving solutions come with potential bias, factual error, and plagiarism risks.”
Previous survey results have suggested that up to 30% of researchers use AI in some capacity when writing manuscripts. This is a trend that has grown substantially in recent years with the emergence of large language models like OpenAI’s ChatGPT.
Although several medical journals, including JAMA and Radiology, have published recommendations on the use of AI in manuscripts, not all journals have followed suit.
According to the new analysis, of 112 radiology journals—80 of which are affiliated with an imaging society—just over 60% have an AI usage policy. Of those, around 58% have specific text and image guidelines pertaining to AI generated content. Those journals tend to have higher impact scores. Journals without such policies do not require that authors disclose the use of AI assistance at all.
The authors implied that the number of journals with specific policies will likely rise in the near future. However, even when new authorship requirements emerge, they will need to be monitored routinely, as the technology will inevitably continue to evolve, the authors suggested.
“We are certain we will face increased capabilities of generative AI systems; thus, even journals with recent AI usage policies will need to update their guidelines,” the authors noted.
Rather than rewriting author policies completely, the authors suggested that guidelines specific to the use of AI in manuscript writing would be more beneficial.
“The main goal of the journals should be easy-to-update dynamic guidelines to continuously keep pace with this rapidly evolving technology. Declaration of all pertinent forms of AI usage with every published article will improve transparency and benefit both journals and readers.”
The study abstract is available here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She joined Innovate Healthcare in 2021 and has since put her unique expertise to use in her editorial role with Health Imaging.