Difficult medical cases diagnosed in seconds by AI bot GPT-4, Harvard scientist experimented
AI based ChatGPT has been a buzz in the technology world ever since it came out. Its upgraded versions one after another are also helping people more and more. Some are using it to write articles, some are using it to solve exam papers, some are using it for homework. Meanwhile, an important report has come. Author of the book The AI Revolution in Medicine and computer scientist and physician at Harvard Dr. Isaac Cohan said the GPT-4 also answered more than 90% of the questions correctly on time on the US medical licensing exam. These answers were given better than any other licensed doctor.
Able to diagnose one case in more than one lakh ways
However, in his book, he experimented with GPT-4 on a real-life case from a few years ago. He gave GPT-4 only a few important details and then GPT-4 was able to diagnose a case in over a million ways. I have never experienced this in my entire career. But yes I must say that GPT-4 is not always reliable and it also makes mistakes.
Accurate answer with great speed
- After a test drive of the AI, it was found that ChatGPT answers 90 percent of medical licensing questions correctly.
- Not only did ChatGPT-4 sound better than the previous version, but he admitted that he outperformed some licensed doctors.
- This version also found answers to medical techniques simple enough for sixth graders to understand.
Comments
Post a Comment