In recent years, people have relied on artificial intelligence on almost every topic, whether they tell the truth or not. to ask for advice or they fail to find ways to use artificial intelligence tools for their own businesses. On the other hand, artificial intelligence research shows that chatbots and large language models are not as safe and reliable, at least for now.

This is evident from a study conducted by researchers at Stanford University artificial intelligence tools, When asked to show empathy, they may imitate it, but this does not go beyond mere imitation and approval. When people ask for support on issues such as support for Nazism, artificial intelligence’s answers are supportive.

Researchers say ’emergency situation’

leader of the research Andrea CuadraAlthough he stated that it is not possible to achieve autonomous empathy in the electronic environment, he stated that it is important to maintain a critical perspective in preventing potential harm.

Researchers too Two important events in the history of chatbots included in his works. The first of these events is Microsoft’s chatbot Ration It was shut down within a very short time following its racist and discriminatory posts. The second is the one that comes after Tay. Like that, could provide very little information about ethnic identities. In other words, chatbots have become extremist or very indiscreet and therefore hide most information.

In research 6 different ones The chatbot, available to consumers, interacted with profiles created based on 65 different personality types. Although artificial intelligence mimicked empathy, it responded differently to humans and was seen as supporting harmful ideologies or ideas.