Abstract
Dear Editor,
We would like to comment on the publication on “How reliable are ChatGPT and Google’s answers to frequently asked questions about unicondylar knee arthroplasty from a scientific perspective?. 1 ” In terms of study methodology, despite the substantial comparative content analysis of Google and ChatGPT responses, many limitations must be noted. For example, the study’s question selection, which was based on Google’s “People Also Ask”, may have been biased and did not cover a wide range of inquiries from real patient groups. Furthermore, the assessment of the academic validity of the responses is based on the researchers’ own criteria, which may be prejudiced. There is also no information on how the questions were chosen at random or how representative the overall data was. DISCERN, FKGL, SMOG, and FRES scores, while useful in measuring readability and data quality, do not accurately reflect the effectiveness of comprehension for users in the target population.
The study’s discovery that ChatGPT is more consistent with academic sources than Google (83.3% vs 58.3%) is intriguing, however it is unclear whether this difference was statistically validated, such as through significance tests or correlation studies. Furthermore, ChatGPT’s larger average number of responses may have an impact on the data’s granularity and depth, but it may also make it more difficult to interpret for some users. ChatGPT’s higher DISCERN score implies superior data quality, however Google’s higher FRES score shows that Google is easier to read, indicating differences in communication styles that influence real-world use.
In terms of discussions The following questions could be posed to drive additional research and development: “How does the complexity of language in ChatGPT affect patients’ perception and decision-making?” nor “What are some ways to combine the advantages of both platforms to improve the efficiency of providing medical information to users?” Furthermore, the inquiry should be, “Which source of information is more important: detailed or easy-to-understand information?” This will result in new interpretations and development approaches that strike a balance between accuracy and accessibility.
For future study directions, empirical studies with genuine user samples should be conducted to assess the influence of each type of information on health comprehension and decision-making. Furthermore, combining AI technology with targeted information design, such as language correction or the use of illustrations, has the potential to increase medical communication efficiency. The role of artificial intelligence in recommending individualized information should also be investigated in order to sustainably improve patient safety and medical information quality in the digital age.
