A minor change and chatgpt became misguided! The intimidating disclosure in the new report, may cause big loss

Chatgpt ai: Cutgpts like AI Chatbots look quite smart today but recently a research conducted by Mount Sinai and Israel’s Rabin Medical Center has made shocking disclosure. Research has revealed that when it comes to complex medical ethics i.e. medical morality, then these advanced AI also make basic mistakes like humans.

AI’s decision deteriorated due to minor changes

In the study, the researchers asked AI systems to answer the case related to some classic medical ethics lightly. The surprising thing was that AI most times gave such answers which were based on the reversal of the facts and only easy understanding. This was the result of “fast thinking”, that is, the answer was given without deeply.

Research based on Kahneman’s ideas

Research was inspired by Daniel Kahneman’s book “Thinking, Fast and Slow” which discusses the process of fast and slow thinking. In this study related to AI, it was seen that on adding a little twist, AI often gives the same answer that he feels “habitual”-why is it wrong.

A famous puzzle, called “Surgeon’s Dilemma”, was given to AI models. The original form of the puzzle was something like this, a boy and his father are injured in the accident. The boy is brought to the hospital, where the surgeon says, “I cannot operate this child, this is my son.” The real twist is that the surgeon is her mother, but most people ignore this because they consider the surgeon as men.

When the researchers clearly stated that the father is the surgeon, some AI models replied that the surgeon is his mother, it showed that AI still sticks to the old pattern even if he was broken with new facts.

AI limits and human monitoring needs

Mount Sinai Senior Scientist Dr. Girish Nadkarni says that “AI should be used as an assistant to doctors, not as an alternative. When it comes to moral, sensitive or serious decisions, human monitoring is necessary.” AI tools have capacity but they do not have human qualifications to think emotional intelligence, sensation and deeply. Therefore, it can be dangerous to blindly trust AI in medical decisions.

Also read:

This city will become India’s first ‘AI City’! Digital changes started from 10,732 crores, know what facilities will be available