Meta ditched its AI-powered celebrity chatbots — featuring images of A-listers and social media influencers like Tom Brady, Charli D’Amelio, MrBeast and Paris Hilton — after the tool failed to generate interest last year when it was launched.
The tech giant — which came under fire this week after its AI-trained software made inaccurate statements about the assassination attempt on former President Donald Trump — reportedly paid an unnamed influencer $5 million to what was supposed to be a two-year pilot program. .
But the overwhelming response from the public, including social media users who found the alternate personas weird and creepy, convinced Meta executives to pull the plug on the project, according to The Information.
Meta CEO Mark Zuckerberg promoted the campaign last year, paying celebrities millions of dollars a year to license their image for a chatbot that would interact with users on Facebook, Instagram and WhatsApp.
Despite celebrities’ online popularity, their AI chatbots struggled to attract large followings.
Snoop Dogg’s AI character, Dungeon Master, only managed 15,000 followers, while the rapper’s real Instagram account has 87.5 million followers.
Kendall Jenner’s AI character “Billie” managed to amass a following of 179,000, which was the most of the 28 AI characters, but still far short of the influencer’s 194 million followers on Instagram.
As of Tuesday, links to these celebrity chatbot accounts on Facebook and Instagram returned messages indicating that the pages were no longer available.
Meta confirmed the removal of chatbots.
A spokesperson told The Information: “We gained a lot of insight from creating chatbots and Meta AI about how people use AI to connect and create in unique ways.”
Earlier this week, Meta said it will launch a new tool called AI Studio that will allow users to create, share and design custom AI chatbots.
AI Studio will allow users to create custom AI characters and will also allow Instagram creators to use AI characters “as an add-on” that can handle common DM questions and story responses , Meta said.
Users can share their AI personas on the social media giant’s various platforms.
The new tool is built with Meta’s Llama 3.1, the biggest version of its mostly free AI models released last week, which is available in multiple languages and has performance metrics that rival paid models from rivals like OpenAI.
ChatGPT maker OpenAI is working on a project codenamed “Strawberry,” the details of which are being kept top secret even within OpenAI, as startups race to show that the kinds of models it offers are capable of providing advanced reasoning capabilities. exclusively Reuters. reported earlier in July.
#Meta #removes #famous #failed #chatbots #users #deeming #scary
Image Source : nypost.com
