Google has taken a huge step in the field of artificial intelligence (AI). A new study shows that AI language models have enough limitations when it comes to medical questions. However, a specialized version may be more effective.
Thus, the American giant has developed a system that has reached a level of accuracy and bias equivalent to that of doctors. This breakthrough paves the way for the use of AI to expand medical capabilities, while highlighting the need for additional updates to achieve a more efficient system. We’ll give you more details here.
Google researchers tested a language model similar to ChatGPT on medical questions
Although artificial intelligence has a high degree of reliability, it is very likely that computer programs will give wrong answers to questions about medicine. Correctly, rapid developments in AI models in the medical field raise concerns about biases incorporated into the answers provided. In this sense, Google researchers presented a language model similar to chat for medical questions.
The results provided insights that the original model incorporated biases into patient reports, which exacerbate health disparities. The study then highlights the need to improve AI models to avoid harmful consequences and ensure accurate results.
Google has developed a specialized version of the Language Model focused on medicine
To overcome issues of bias and inaccuracy, Google has developed a specialized version of the language model that focuses on medicine. This more detailed version of the form is designed to better meet the specific needs of healthcare professionals and patients.
Indeed, the researchers worked to integrate expert medical knowledge to improve its accuracy and reduce false answers. The results of the study showed that this artificial intelligence specialized in medicine obtained similar performance to that of doctors. As estimated by the Physicians’ Committee, 92.6% of the responses produced by the medicine-centered form were consistent with the scientific consensus, a number close to the 92.9% reported by the physicians themselves.
Thus, this breakthrough is promising in the medical field, as it shows that AI can be used to help doctors make their decisions and provide reliable and accurate information. However, it is important to note that this specialized AI still exists support tool It should not be considered a substitute for the knowledge and experience of physicians. Complex medical decisions require in-depth human evaluation, taking into account each patient’s individual context.
Challenges and prospects for using artificial intelligence in medicine
Despite the progress made, more work is needed for the effective use of AI in medicine. Correctly, several clinicians have argued that the non-specialist model only conforms to the scientific consensus in 61.9% of cases. On the other hand, the specialized model achieved a rate of 92.6%, in line with the physicians.
Furthermore, medicine is a constantly changing field, with new research, technological developments, and medical discoveries occurring regularly. It is essential that medical AI models are able to adapt and integrate this new knowledge to maintain relevance and accuracy.
source : sky news
“Hardcore beer fanatic. Falls down a lot. Professional coffee fan. Music ninja.”