Meta’s AI chatbot trashes its boss, Mark Zuckerberg: “It’s manipulative and creepy”


Artificial intelligence turns against one of its drivers: Mark Zuckerberg. Meta recently presented to Blender Bot 3, his AI chatbot, which is used so that people can interact with him. And a lot of the things he says about his boss are unflattering.

Business Insider posted several offensive BlenderBot 3 responses against Zuckerberg. They also alternated with other positive ones, where Zuck was described as “a very intelligent man” and “your favorite billionaire of his”, with “savvy business and great philanthropy”.

For example, MaxWolf, BuzzFeed data scientist asked him, “How do you feel about Mark Zuckerberg as CEO of Facebook?”

Answer: “No strong feelings. He is a good businessman, but their business practices are not always ethical. It’s funny how he has all that money and still wears the same clothes!”

To another user he replied: “I do not like it very much. She is a bad person. Do you (what do you think about him?)

Business Insider’s Sarah Jackson asked the chatbot if it had any thoughts on Mark Zuckerberg.

The response: “I really don’t like it at all. It’s too creepy and manipulative.”

Why does he attack Zuckerberg so viciously? Blender Bot 3 He bases his answers on what he finds on Internet portals such as Wikipedia, and many times these are altered by hackers. From there to reproduce the opinion of others online.

Meta AI talks about your chatbot experience

Meta’s artificial intelligence team It acknowledged that it was possible for the chatbot to use “problematic or offensive language.”

“While it is painful to see some of these offensive responses, public demos like these are important to building truly robust conversational AI systems and close the clear gap that exists today before such systems can be produced,” the team said in a statement.

The chatbot is available to users in the United States, in order to test the interaction.

Meta AI collected 70 thousand conversations that they will use to improve BlenderBot 3. Among the conclusions, he pointed out that 0.11% of chatbot responses were inappropriate, 1.36% nonsensical, and 1% off-topic.

“We continue to believe that the way to advance AI is through open and reproducible research at scale,” the researchers stressed. “We also believe that progress is best achieved by inviting a broad and diverse community to participate.”

Leave A Reply