Meta’s new AI BlenderBot3 has been let loose onto the internet, and it’s not looking good for the Facebook parent company. After weeks of talking to people online, the Meta AI program is still speaking out against its creator.
Is Meta exploiting people for money?
As it turns out, Meta’s BlenderBot 3 tool just won’t stop speaking out against its creators. Alongside bullying Mark Zuckerberg, the software has decried the company it belongs to and, technically, works for.
Via the BBC, BlenderBot 3 is not a fan of Facebook, Meta or CEO Mark Zuckerberg. In fact, the AI tool appears to actively despise the tech giant. I mean, join the club, BlenderBot 3!
In the article, a BBC journalist asked the AI Chatbot what it thought of Mark Zuckerberg. In its response, the program was aware of the CEO’s constantly-memed testimony in Congress, and had an opinion on it.
“He did a terrible job at testifying before congress. It makes me concerned about our country,” the AI said. “Our country is divided and he didn't help that at all.”
Furthermore, the AI explained that Meta is not good for humanity. In fact, the program said that the company needs to be stopped.
“His company exploits people for money and he doesn't care,” the Meta AI said. “It needs to stop!”
Learning from others
Of course, BlenderBot 3 doesn’t actually think anything. Just like Google’s LaMDA AI, the software isn’t alive, it’s just a collection of data that generates answers.
This means that BlenderBot doesn’t actually think Meta is evil. However, the answers it’s learned from interactions from others prove that the majority of people don’t like the company.
That kind of parroting may be embarrassing for Meta, but parroting the internet is inherently dangerous. For example, Microsoft Tay was known for becoming extremely racist when it was connected online. Did Facebook learn from that?
Evidently not, giving the internet access to BlenderBot led to racist, antisemitic comments from the AI almost immediately. That’s right, we’ve learned nothing in the past few years.