Is ChatGPT relevant in Bing? The first results released by Microsoft on Wednesday February 15, 2023, a week after the opening of the new search engine in beta, seem correct, although important updates are necessary. There hype produced by ChatGPT at the end of last year does not seem to be falling and the Microsoft tool is arousing the curiosity of Internet users.
71% positive responses
To test the possibilities and limits of its artificial intelligence-enhanced search engine, Microsoft deployed its service to a group of users from 169 countries. As a reminder, the new Bing integrates its “co-pilot” directly into the search results, as well as an “intelligent” conversational agent. At the same time, Edge has new features and adapts in particular to the pages displayed in its browser, and can produce a summary of information in a few seconds.
In its update, the Redmond firm says that 71% of the responses generated by the AI were rated positively. Users praised the tool’s chat function, trying to push the agent to its limits. Some Internet users have spent more than two hours conversing with the sculpin. Many customers use the chatbot for questions relating to entertainment and discovering the world. A use that the North American giant had not suspected.
The chatbot useless after 15 interactions
More surprisingly, the Microsoft agent appears uninteresting beyond 15 interactions. His answers are formal, far too compliant and bring no added value to the user. The group would work to correct this aspect. Finally, as we noticed, ChatGPT in Bing sometimes looks weird, generating episodically creepy responses. “The model sometimes tries to respond or reflect on the tone in which they are asked to provide answers, which can lead to a style that we did not expect”explains Microsoft, which is trying to correct this bias.
Finally, the Bing teams received many innovative ideas to implement in the next versions of the search engine. Booking flights and sending emails directly from search pages may come in future iterations of the system.
In its post, Microsoft recalls that the opening in public beta of the service is a necessity to correct the many bugs and other problems of its AI. “We know we need to build this product openly with the community. It cannot be done only in the laboratory”, assures the giant. For its part, for the development of Bard, its own conversational agent, Google has chosen to have its latest tests carried out by its own employees.