Meta’s new artificial intelligence trashed Mark Zuckerberg and mocked his clothes

In a round of questions, the BlenderBot answered without a filter about the CEO of the company.

Many of the current chatbots are like an intelligent menu where two replicas appear for each option. However, those who develop these solutions are looking for ways to take them to the next level. goal, of Mark Zuckerbergpoints in that direction and has just presented the third generation of BlenderBot. The problem, of course, is that this software lacks feedback inhibitions.

To improve its service, Meta indicated that it completed the training with a new database consisting of more than 20,000 conversations with people on more than 1,000 topics, to be able to interact in a “smarter, safer and more useful” way.

However, in the first open round he surprised everyone with his sincere answers. Thus, the data scientist of BuzzFeed, Max Woolfasked him what he thought of Mark Zuckerberg as CEO of Facebook.

“No strong feelings. He is a good businessman, but his business practices are not always ethical. It’s funny how I have all this money and still wear the same clothes.“Commented the chatbot shamelessly.

In view of his verbiage, a Twitter user asked him a trick question. He told her if he believed that the metaverse was over. And the answer left everyone speechless:

“So it seems, though Facebook still has a lot of money invested and will probably continue to do so for years to come.”

One of BlenderBot 3’s abilities is learning as you go and improving the quality of its service. That way, when one of your answers is “unsatisfactory”, the platform change the focus to refine your conversational model.

Thus, when a new user asked him about Zuckerberg again, he rehearsed a positive answer “He is great and a very intelligent man. He’s my favorite billionaire!”

How is BlenderBot, the Meta chatbot

At this stage of testing, the chatbot only works with the English language and in the United States. Meta’s artificial intelligence lab explains that it has been designed to learn from interactions of the users.

To operate, it combines two recently developed machine learning techniques, SeeKeR and Director, to build conversational models that can learn from interactions and feedback.

In this sense, the developer has commented that Director is capable of generating responses based on two mechanisms: language modeling and classification.

The first provides this chatbot with the most relevant answers, based on the data collected during its training.

Subsequently, the classification mechanism helps you distinguish between a correct and incorrect response based on human feedback. Thus, to generate a response phrase, both mechanisms must agree and complement each other.

The original version of BlenderBot appeared two years ago. Back then he had the ability to combine “empathy and knowledge” in conversations.

A year later version 2 arrived, with a long-term memory capacity based on artificial intelligence to maintain more sophisticated conversations on any topic.

By Editor

Leave a Reply