Bing, Microsoft’s search engine, is purportedly leveraging the popularity of the AI chatbot ChatGPT by incorporating it inside the search engine. This was disclosed when a new, AI-assisted version of Bing emerged and then vanished today. Owen Yin, a student and designer, reported seeing the “new Bing” on Twitter and gave further information on the integration in a Medium blog post.
According to Yin, the new Bing chatbot could not only answer questions but also ask them in a conversational tone. Furthermore, the chatbot was able to cite its sources, which is a significant feature since the absence of source information in language models such as ChatGPT might render them untrustworthy. Screenshots of the new Bing reveal a new “chat” option next to “search” on the menu bar. When clicked, a chat interface appears with the text “Welcome to the new Bing: Your AI-powered response engine.”
Despite the allure of AI-assisted search engines, there are worries concerning the accuracy of information supplied by AI language models. AI language models are known to “hallucinate” and provide erroneous information on a regular basis. This begs the issue of whether the information provided by AI language models in search engines is potentially hazardous. If these systems are allowed to operate without adequate protection, they may deliver hazardous medical advice, racist or biased renditions of history, and other potentially deadly information.