פוסט זה זמין גם ב:
עברית
Written by Gabby Leonard
ChatGPT is an artificial intelligence software that is being explored for scientific writing. While there is great potential for utilizing this in the medical field, it should not be used in isolation and requires human oversight for both concept proof and personalization.
Why does this matter?
Scientists are optimistic about capabilities for ChatGPT to aid with writing articles, extract and summarize data, provide suggestions for editing/formatting as well as to generate drafts for publications. Concerns remain over how to most appropriately implement this into an emergency department setting.
Are robots taking over the world?
Chat Generative Pre-trained Transformer (ChatGPT) is a type of artificial intelligence (AI) software that uses language models to create written responses to various keywords or phrases. Apprehension persists over authenticity and inaccuracies of these automated responses.
Algorithms are designed to understand and respond to natural language and answer with appropriate responses that can be pre-fabricated or entirely novel. While this AI is unable to generate new ideas, it instead focuses on organizing and developing existing ideas and thus lacks the creativity and critical thinking of humans.
Detailed analysis of utilization for Emergency Medicine processes needs to be completed before implementing this software for anything directly related to patient care (discharge summaries, note-writing, etc.).
Potential identifiable uses in medicine include:
- ICU monitoring of multiple patient parameters at one time, including vital signs, laboratory values and test results, as well as response to treatment
- Design & adherence notifications around patient-centered protocols
- Generating clinical notes, including progress notes and discharge summaries
- Translation for language barriers.
Ultimately AI may be used to aid with scholarly writing; however, the need for human supervision and personalization remains constant.
*The authors asked ChatGPT to review and approve the article’s submission, to which it responded, “It’s important to keep in mind that ChatGPT should always be used in combination with the expertise and judgement of human experts and its output should be validated before it is used in clinical practice.”
Editor’s note: I recently asked ChatGPT about management of intermediate risk PE. It gave a very confident, articulate but completely wrong answer. ~Clay Smith
Source
Can artificial intelligence help for scientific writing? Crit Care. 2023 Feb 25;27(1):75. doi: 10.1186/s13054-023-04380-2.