Microsoft Bing AI Chatbot has just been updated with several important upgrades and improvement, including support for LaTeX markup, a reduction in end-of-conversation triggers, and news grounding as well as fixing problems with the chatbot triggering duplicate searches and reducing errors that would lead to the chatbot not giving a response.
According to a recent blog post from Microsoft, LaTeX markup is set to enable Bing’s chatbot to better display complex mathematical expressions, meaning Bing’s bot will now be much better at helping users learn mathematical concepts or write research papers, among other things.
Another key update is a reduction in end-of-conversation triggers. Sometimes, Bing’s chatbot could suddenly end a conversation for little reason, and this update is aimed at making sure that happens less.
News grounding, according to Microsoft, is another key feature of this latest update to Bing’s chatbot. It’s unclear exactly what the details are here, but according to Microsoft, now Bing’s chatbot will give you better answers when it comes to asking questions about news or current events. Though, Microsoft does note that users should continue to expect the company to make further improvements to news-related queries depending on feedback.
Lastly, a couple of bugs with Bing’s chatbot have been fixed with this new update. For one, Microsoft has updated the chatbot to trigger fewer unnecessary duplicate searches. Also, Microsoft has worked on minimizing issues where the chatbot would not return any response to a query for no apparent reason. Once again, it’s unclear exactly how Microsoft accomplished all that, but you can try out the chatbot and see for yourself how well it works.
All of the above are fairly meaty upgrades to Bing’s chatbot, and it’s good to see Microsoft continuously update its AI chatbot. As this update is live now, users can load up Bing to start enjoying the latest version of its AI chatbot.
When you ask a multiple question like what is something named and over what dates, chatbot will only record and answer the first part without combining it with the second part. This is unsatisfactory and it need to take into account all of the words in the question.