I recently read the article published titled “Assessing the performance of ChatGPT in answering questions regarding cirrhosis and hepatocellular carcin” [1]. The authors have conducted a thorough evaluation of the chat generative pretrained transformer (ChatGPT) model, identifying its strengths and limitations in providing knowledge, management, and emotional support to patients and physicians. As a researcher working with the latest generative pre-trained transformer 4 (GPT-4) model, I would like to present some points that may warrant a follow-up study, focusing on the potential advantages of GPT-4 over its predecessor [2].

GPT-4, as the successor to ChatGPT, offers a range of improvements and features that make it more advanced and versatile than its predecessor. One of the key factors that distinguish GPT-4 is its larger scale, trained on an even more extensive dataset, which allows it to learn from a broader range of knowledge sources and contexts. This enhanced understanding enables GPT-4 to generate more comprehensive and accurate information, particularly in complex domains like medical decision-making. Moreover, GPT-4’s improved context awareness results in better handling of long-form text, reducing the chances of generating irrelevant or repetitive content. Additionally, GPT-4 can accept a prompt that combines text and images and, in contrast to the text-only setting, allows the user to specify any language or vision task which would be helpful for patients. GPT-4’s enhanced understanding of context may address some of the limitations identified in ChatGPT’s responses, particularly in the domains of diagnosis, preventive medicine, decision-making cut-offs, and treatment durations [3]. This improved comprehension could lead to more comprehensive and accurate information, helping patients and physicians make better-informed decisions. GPT-4 can adapt to regional guidelines by training on specific datasets [4]. This feature can potentially offer more geographically relevant information, addressing the need for knowledge of regional variations in hepatocellular carcinoma (HCC) screening criteria and other guidelines, as pointed out in the article. GPT-4’s empathetic response capabilities can be fine-tuned, allowing for better emotional support for patients and caregivers. This improvement can significantly enhance the model’s utility as a support tool, further bridging the gap between artificial intelligence (AI)-generated advice and human compassion [5].

Given these advancements in GPT-4, a follow-up study examining its performance in the context of cirrhosis and HCC management may be needed as it would not only build upon the existing work but also highlight the potential of GPT-4 as an adjunct informational tool in the medical field [1,2,5].

Citations to this article as recorded by