Microsoft is limiting how extensively folks can converse with its Bing AI chatbot, following media protection of the bot going off the rails throughout lengthy exchanges. 

Bing Chat will now reply to as much as 5 questions or statements in a row for every dialog, after which customers will probably be prompted to start out a brand new matter, the corporate stated in a weblog put up Friday. Customers can even be restricted to 50 complete replies per day. 

The restrictions are supposed to preserve conversations from getting bizarre. Microsoft stated lengthy discussions “can confuse the underlying chat mannequin.” 

On Wednesday the corporate had stated it was working to repair issues with Bing, launched simply over per week earlier than, together with factual errors and odd exchanges. Weird responses reported on-line have included Bing telling a New York Occasions columnist to desert his marriage for the chatbot, and the AI demanding an apology from a Reddit person over whether or not we’re within the yr 2022 or 2023.

The chatbot’s responses have additionally included factual errors. Microsoft stated on Wednesday that it was tweaking the AI mannequin to quadruple the quantity of information from which it could possibly supply solutions. The corporate stated it will additionally give customers extra management over whether or not they need exact solutions, that are sourced from Microsoft’s proprietary Bing AI know-how or extra “inventive” responses that use OpenAI’s ChatGPT tech.

Bing’s AI chat performance remains to be in beta testing, with potential customers on a wait record for entry. With the device, Microsoft hopes to get a head begin on what some say would be the subsequent revolution in web search. 

The ChatGPT know-how made an enormous splash when it launched in November, however OpenAI itself has warned of potential pitfalls, and Microsoft has acknowledged limitations with AI. Regardless of AI’s spectacular qualities, considerations have been raised about synthetic intelligence getting used for nefarious functions like spreading misinformation and churning out phishing emails.

Learn extra: Generative AI Instruments Like ChatGPT and Dall-E Are All over the place: What You Have to Know

With Bing’s AI capabilities, Microsoft would additionally prefer to get a leap on search powerhouse Google, which introduced its personal AI chat mannequin, Bard, final week. Bard has had its personal issues with factual errors, fumbling a response throughout its first public demo.

In its Friday weblog put up, Microsoft advised the brand new AI chat restrictions are primarily based on info gleaned from the beta check.

“Our information has proven that the overwhelming majority of you discover the solutions you are searching for inside 5 turns and that solely ~1% of chat conversations have 50+ messages,” it stated. “As we proceed to get your suggestions, we’ll discover increasing the caps on chat classes to additional improve search and discovery experiences.” 

Editors’ word: CNET is utilizing an AI engine to create some private finance explainers which are edited and fact-checked by our editors. For extra, see this put up.