Beneath its harmless appearance, the Furby plush has long been a terror in homes. Activation in the middle of the night, impossible to turn off or endless arguments: any parent who has offered the electronic toy to his son knows the ordeal it causes. But a programmer had the idea to connect a Furby to ChatGPT, with a terrifying result, says the British newspaper The Independent.
A cute and cuddly look
Connected to OpenAI artificial intelligence using a Raspberry Pi microcomputer, the stuffed animal made disturbing comments. When asked about a hidden plot to take over the world, Furby detailed a Machiavellian plan.
“The Furby’s plan to take over the world is to infiltrate homes with their cute and cuddly appearance,” the AI-powered plush begins. He then claims that the toys will use “advanced technology to manipulate and control their owners.”
Looking at how ChatGPT works, Furby’s quote is not surprising. Fueled by web content, artificial intelligence has brought these arguments back from Facebook. The words of the stuffed animal repeat word for word an intriguing 2017 publication by the Futurism medium, says The Independent.
The only problem is that the post has been cached ever since, making it impossible to get the context. On Wednesday, OpenAI said in a blog post that its teams continue to work on the false information presented by its tool. Without a specific context, ChatGPT would no longer be able to take this appointment into account in the future.
Source: BFM TV
