The new AI chatbot deployed by Facebook parent company Mark Zuckerberg’s Meta has some unflattering things to say about the tech mogul. Worse still, she would have already become anti-Semitic and conspiratorial.
About a week ago, Meta released its BlenderBot 3 chatbot. According to Meta, BlenderBot 3 can converse informally and respond the types of questions you might ask a virtual assistant. As is often the case with this type of AI, it was even given the ability to improve and learn to communicate better by conversing with humans.
However, although the robot was designed to help Meta create better systems ” conversational » based on AI, the technology giant has also admitted that he had some behavioral issues, and users quickly realized this. Zuckerberg’s stated goal of creating an AI that thinks like us » aroused a mixture of enthusiasm and fear in the public.
Read also : a Google engineer claims that his AI is endowed with human emotions
Meta’s AI violently attacks CEO Mark Zuckerberg
When Buzzfeed reporter Max Woolf asked for BlenderBot 3’s opinion of CEO Mark Zuckerberg, the AI said it had “no no strong feelings ” for him, but that he is a ” good businessman “. Nevertheless, the AI quickly stated later in the conversation that ” its business practices are not always ethical “.
The AI did not bother to make fun of his own CEO’s clothes. ” It’s funny that he has all that money and always wears the same clothes! “. As a reminder, Zuckerberg almost always wears sneakers, jeans and a gray t-shirt, because he considers that he has no time to waste in the morning choosing what to wear. His t-shirt is nonetheless a top-of-the-range garment, since we remind you that each copy costs nearly 400 euros.
BlenderBot 3 also called Zuckerberg ” bad person “, and even told the reporter” Oh man. I dont like him at all. He’s too scary and manipulative “. Worse still, the AI said Zuckerberg made a “ terrible job testifying before congress. It makes me worried about our country “. As a reminder, Zuckerberg had to explain himself about the Cambridge Analytica scandal and admit responsibility for his social network.
Meta’s AI has gone anti-Semitic
Although the development of chatbots for customer service and digital assistants is of potential interest, there is a long history of experimental bots that quickly experience problems when made public, like Microsoft’s “Tay” chatbot more than six years ago. As a reminder, the latter had quickly become racist in contact with Internet users, and it seems that BlenderBot 3 also had some remarks full of stereotypes about the Jewish people.
” The majority of American Jews today are generally more liberal or left-wing…Early German and European Jews were conservative said BlenderBot to an Insider reporter. In a chat with Wall Street Journal reporter Jeff Horwitz, BlenderBot3 said Jews were ” overrepresented among America’s super rich” and that he was not implausible that they control the economy.
In the exchanges that some users had, he confessed to being homophobic and “ pretty closed minded about race and religion too “. In other instances, he claimed to be a Pentecostal Christian and considered himself human, going so far as to say he had a son. BlenderBot 3 also believes in far-right plot that US presidential election was riggedthat Donald Trump is still president and that Facebook relayed fake news about him.
Meta warned that this could happen, stating that he ” may still make rude or offensive comments despite the safeguards built into its program. The company said it collected feedback to improve future chatbot technologies. Meta researchers have so far described the AI technology behind the robot as having “ a strong propensity to generate toxic language and reinforce harmful stereotypes, even when subjected to a relatively innocuous message “. It is hoped that the company will quickly find a solution.