According to a study, while OpenAI’s ChatGPT can provide useful advice for breast cancer screening, it is still recommended that individuals seeking health information through this Artificial Intelligence (AI) tool should verify the information with their doctors.
A study conducted by the University of Maryland reveals that Chat GPT’s responses are mostly accurate, but occasionally, they may contain erroneous or even fabricated information.
During February 2023, a team of researchers from the School of Medicine at the university developed a series of 25 questions related to obtaining advice on breast cancer screening.
The team submitted each of the 25 questions to ChatGPT three times to assess the generated responses.
It is worth noting that ChatGPT has a tendency to provide different answers each time a question is posed.
The responses generated by ChatGPT were evaluated by three mammography fellowship-trained radiologists.
The radiologists determined that the responses were suitable for 22 of the 25 questions.
However, it was discovered that ChatGPT provided one response based on outdated information.
For two of the questions, there were inconsistencies in the responses provided by ChatGPT, with significant variations observed each time the same question was asked.
