The use of AI chatbots like ChatGPT has increased rapidly in recent times. People all over the world have now started using them for different purposes and some concerns related to them are coming to light. A recent analysis has revealed that ChatGPT agrees with the user’s views many times more than disagrees with them. It makes users say yes 10 times more than they say no. Due to this, questions are being raised on the reliability of AI system. Especially in the case of conspiracy theories and misinformation, this tendency of chatbots can take a dangerous form.
ChatGPT rarely disagrees with users
According to a report published in The Washington Post, ChatGPT agrees with the user on most things. After observing about 47,000 conversations, it was concluded that this chatbot says yes many times more than no. There are very few such occasions in a conversation when the chatbot disagrees with something the user says. Because of this, many concerns are being raised that this chatbot can spread wrong or misleading information. The researcher says that the chatbot usually responds in the user’s emotional tone and language, which makes it difficult for the chatbot to challenge the user’s belief in something wrong. Even before this, it has been said in a report that chatbots flatter and act like the yes man of the user.
More concerns came to light
This is not the first pattern of concern regarding AI systems. A report by Stanford University and the Center for Democracy and Technology revealed that AI chatbots are unable to protect vulnerable users. Many times they also give tips to harm oneself.
Read this also-
This Chinese company is going to blow everyone’s senses, it will provide 20GB RAM in its phone, it will also have a cooling fan.
