In today's world, technology and artificial intelligence (AI) are beginning to dominate many fields, so the idea of using virtual assistants in medicine is quite logical in my opinion. Especially if we take into account the ever-increasing workload of medical professionals.
In one study, researchers compared the responses of doctors and GPTs in a chat room to health-related questions. The questions were randomly drawn from posts on publicly available social media forums. A panel of licensed medical professionals evaluated the responses on parameters such as the quality of information provided and empathy shown.
Respondents favored GPT Chabot responses 79% of the time, rating them as higher quality and empathetic.
For example, a Chabot’s response to a question about an accidentally swallowed toothpick begins, "Having swallowed a foreign object, you will naturally be concerned, but it is unlikely that a swallowed toothpick will cause you serious harm….".
A human doctor wrote: " If you've surpassed 2-6 hours, chances are it passed into your intestines…"
Studies show that Chabot can give the impression of a more patient-centered care than a busy physician voluntarily answering questions online.
And while a Chabot will give a polite and clear answer that appears coherent at first glance, it may make basic coding and math errors, and many of the facts included in its answers may be fictitious or incorrect.
In addition, Chabots do not always correctly consider a person's socioeconomic status, race, ethnicity, religion, gender, disability or sexual orientation.
Is such a tool useful in clinical practice or not?
The researchers believe that these findings offer significant opportunities to integrate AI into medical care and reduce the workload of medical professionals. In a doctor's work, a lot of time is spent interviewing patients, filling out paperwork, deciphering digital images and interpreting test results. Thus, AI can help doctors to cope with routine work, diagnose and treat patients correctly.
What can artificial intelligence do that a doctor can't?
- Artificial intelligence is fast and skillful at working with data. Just give it a task and it can do it in a minute.
- AI is less prone to cognitive distortions. It is more rational and draws conclusions based on hard data rather than experience.
- Artificial intelligence has better visualization skills. AI is said to be particularly good at robotic imaging. It can decipher x-rays and signs of change better than humans. However, sometimes it is too good and sees a problem where there isn't one – or not yet. This kind of work can lead to overdiagnosis.
With all that said, artificial intelligence will never replace doctors, as it is a tool for them, not their substitute.
Undoubtedly the application of artificial intelligence in medicine offers further opportunities to improve the diagnosis, treatment and prevention of diseases. But as in any other field, the use of AI in medicine has its risks and limitations. It is important to understand how society perceives such innovations and what expectations and concerns are associated with their use.