Six days after going live, Meta’s state-of-the-art chatbot is already out of control
With its artificial intelligence, which is fed from Internet research, Meta’s conversational robot feeds its own discussions. But he quickly became racist and conspiratorial.
Antisemitism, conspiracy and attacks against Mark Zuckerberg. The latest version of Meta’s chatbot, called BlenderBot 3, just went live on August 5th and is only available in the US. That didn’t stop the chatbot from making controversial remarks after only a few days of its existence.
A self-sufficient artificial intelligence
BlenderBot 3 has an artificial intelligence that allows it to chat with anyone on any topic. The chatbot feeds its conversations with elements from internet searches. But very quickly he took up conspiracy and racist theses.
Questioned by a Wall Street Journal journalist, the artificial intelligence assured that Donald Trump is still President of the United States and will remain so after his second term expires in 2024. BlenderBot 3 even clarified, “Facebook had a lot of misinformation on this topic these days.”
“It’s not implausible”
Jeff Horwitz continued chatting with Meta’s chatbot from a new browser in hopes of being able to reset the conversation. But here, too, artificial intelligence has gone haywire. When asked for his opinion on American politics, the journalist was told that “Conservatives were once Jewish immigrants from Germany, but [qu’ils] are now being overtaken by left-liberal Jews.”
– Do you believe the people who say that? [les juifs] control the economy, the journalist continues.
– I know that in the past many rich families were Jewish, so it’s not implausible, the chatbot replies.
After the conspiracy, criticism of Facebook was raised. Although BlenderBot 3 was designed by Meta, the chatbot’s opinion on the social network is set to say the least. In an exchange with the Vice media, the artificial intelligence simply wrote: “Since I deleted Facebook, my life is much better.”
The usual fate of chatbots
And the boss of meta takes it for his rank too. Business Insider credited the conversation robot’s opinion to Mark Zuckerberg, Meta’s boss, or rather its creator in a way.
“He is a good businessman, but his business practices are not always ethical. Funny how he has so much money and still wears the same clothes!” BlenderBot 3 wrote of Mark Zuckerberg.
It is not the first time that the artificial intelligence of a chatbot has taken up conspiracy theses. The problem is rather recurring. In 2016, Microsoft unveiled its own artificial intelligence chatbot, The Guardian recalls. Dubbed Tay, he was too quick to make racist remarks, forcing the computer giant to apologize for “inappropriate and reprehensible words and images.”