According to a new study of the BBC, the world’s most popular artificial intelligence chat robot, including Chatgpt, misrepresented the news.
The BBC asked Chatgpt, Copilot, Gemini, and Perplexity, the first four, to summarize the 100 news in the news organization and then rated each answer to determine how accurate the answers of artificial intelligence were.
In the study, 51 %of the artificial intelligence answers given to the questions about the news contain important problems ”and“ 19 %of the artificial intelligence answers referring to the BBC content contained false factual expressions, numbers and dates ”.
Among the examples given for a tremendous amount of misconduct, Gemini says that “NHS does not suggest electronic cigarettes by helping to quit smoking” and “Chatgpt and Copilot, Rishi altar and Nicola Sturgeon say that they are still on duty even after leaving” is located.
Can he not separate artificial intelligence, ideas and truth?
However, in addition to these false information, there is another significant finding. In the study of the news organization, artificial intelligence was “that it has difficulty distinguish between ideas and reality, behaved editorially and often failed to include the basic context”.
Considering the problems that are now frequently seen in news summary tools, which are currently frequently seen in news summary tools, which also lead to the temporary lifting of Apple Intelligence in iOS 18.3, it can be said that this result is not very surprising. Nevertheless, this study reminds you that you should not fully trust the summary of artificial intelligence.
According to the result of the BBC study, “Microsoft’s Copilot and Google’s Gemini, OpenAI’s Chatgpt and Perplexity have more important problems.”
While the study does not provide more information about the subject, it once again confirms the doubts about artificial intelligence summarizing tools. Although artificial intelligence and large language models are developing rapidly, there are many problems that are still not perfected with this rapid development. While errors are eliminated at a speed similar to developments, you should approach the things you want to summarize from artificial intelligence vehicles for now.
