Android updates

BlenderBot 3: Facebook’s chatbot criticizes Mark Zuckerberg and makes problematic comments

For a few weeks now, the web has been passionate about the releases of BlenderBot 3 , the chatbot developed by the big heads of Meta/Facebook. It must be said that the conversational AI does not have its language in the processor. At the beginning of August, BlenderBot 3 severely tackled Zuckerberg, the CEO of Meta, declaring that he had no ” a good impression “ from the promoter of the Metavers. “He is an efficient businessman” add AI, “but its practices in the business world are not always ethical. It’s funny that he has all that money and still wears the same clothes! ». Breaking… BlenderBot 3 is also inconstant since it will make laudatory remarks on Zuckerberg a few days later.

Another detail makes people cringe: BlenderBot 3 seems to have the same flaws as the chatbot Tay of Microsoft, who had the annoying habit of integrating a little too quickly the most problematic “opinions” of his interlocutors, to the point of making deeply racist or sexist remarks. The Redmond firm even had to withdraw its public chatbot in the face of controversy. Unfortunately, BlenderBot 3 does not do much better in the end, as evidenced by several edifying discussions (see below):

Meta claims to have succeeded in reducing the production of problematic comments from BlenderBot 3 by 90%. Since all conversational AI chatbots are known to sometimes mimic and generate dangerous, biased, or offensive remarks, we conducted large-scale studies, co-hosted workshops, and developed new techniques to provide safeguards for BlenderBot 3. » specified a Meta press release not long ago, while specifying that despite these efforts, the chatbot continued to spout nonsense (fortunately much less frequently).

Leave a Reply

Your email address will not be published. Required fields are marked *