Microsoft announces preliminary test results of Bing's integration of ChatGPT: search content is approved by 71% of testers
According to news on February 16, Microsoft's move to integrate artificial intelligence chat robots into its Bing search engine and Edge browser is showing signs of success, with 71% of testers approving Bing search results optimized by artificial intelligence.
Microsoft conducted a week-long test of the new artificial intelligence features of Bing search engine on users from 169 countries, and released preliminary results on Wednesday local time.
Microsoft said 71% of testers approved the search results optimized by artificial intelligence. The artificial intelligence chat robot function has also proved to be very popular and can deepen user participation. People not only query content for specific purposes, but also use Bing search, which has an artificial intelligence chat function, to "discover the world in a broader way and engage in social entertainment", the company said.
The new features integrated by Bing are the result of Microsoft's integration of OpenAI artificial intelligence chat robot ChatGP technology. As a start-up in the field of artificial intelligence, OpenAI was supported by Microsoft early in its operation, and Microsoft recently announced that it would invest $10 billion in the company and deepen the cooperation between the two sides. Microsoft's design of integrating chatbots into search engines is designed to enable search engines to give more complete answers in a more interactive environment.
Microsoft says there is a lot of room for improvement in the new features. For query functions that require a high degree of accuracy, such as the financial statements of listed companies, Microsoft will send four times as much basic data to the model responsible for generating answers. Microsoft also found that the artificial intelligence chat function in the search engine did not perform well in longer conversations with more than 15 questions. "the results given by Bing may become repetitive or give answers that are not necessarily useful or do not match the tone of our design according to special prompts," the public name said. "
Recently, researchers and other users have posted some unusual interactions with Bing chatbots, in which artificial intelligence chatbots even show a threatening or controversial tone, or give content that is clearly controversial with the basic facts. Microsoft acknowledges that someone is testing marginal use cases and says these behaviors help improve the product.
"We know we have to work with the community to build this system openly, not just behind closed doors in the lab," Microsoft said. "
