Celebrities could make hundreds of millions of dollars a year if they allowed companies to use their images to develop AI-powered virtual assistants. This seems to be demonstrated by the contract between Meta and the celebrities behind its recent AI chatbots. A report published last week claimed that Meta will likely pay each of the stars up to $5 million for six hours of work. The report sparked outrage and Meta came under sharp criticism. Some argued that the people (often outsourced workers) who work to make AI systems less toxic also deserve good pay.
Last month, at the 2023 edition of its annual Meta Connect event, Meta introduced a series of 28 AI chatbots, interactive assistants based on real celebrities, including Tom Brady, Paris Hilton, Kylie Jenner, Snoop Dogg, MrBeast and Charli D’ Amelio. Each AI chatbot interacts with users and provides advice on the star’s specific area of expertise. (For example, a chatbot imitating Tom Brady’s character “Bru” can give the user advice about American football; Charli D’Amelio’s “Coco” can give advice about dancing; and Kylie Jenner’s “Billie” can play the role of Ride-or-Die companion.)
The chatbots are controlled by Llama 2, a Large Language Model (LLM) developed by Meta. They will be available across Meta’s family of social and messaging apps (Instagram, WhatsApp, Messenger and Facebook) and could be expanded to virtual reality experiences on Meta’s Quest 3 mixed reality headset. No details were given during the presentation about Meta’s business model for these chatbots or the celebrity payment models involved, but it does hint at a new type of monetization opportunity for creators. This could bring very big profits to the creators.
In a report published last week, The Information claimed that Meta could pay a celebrity up to $5 million for six hours of work in a studio to leverage her image as an AI-powered virtual assistant. Note that Meta has also recruited YouTubers like MrBeast, the most subscribed person on YouTube. According to the report, Meta was initially willing to pay more than $1 million to use the stars’ likeness, but paid more for the big names. The report does not specify who the $5 million was paid to, but mainly refers to the creators.
The report quoted a person familiar with the matter as saying that other deals with high-profile designers were in a similar price range and that the deals were non-exclusive, meaning the stars could sell their image to other AI products if they wanted . Currently, AI assistants are only available on a text basis. Bots answer questions and have real-time human conversations with the user based on the character’s personality. However, Meta’s demo video included clips of celebrities speaking as AI assistants.
Advances in AI allow us to create different AI characters to help us complete various tasks. It won’t just be about responding to requests. “This is done for entertainment and to help you connect with the people around you,” Meta CEO Mark Zuckerberg told developers at the annual Meta Connect conference in Menlo Park, California. Zuckerberg said there is a huge demand for celebrity AI assistants. However, he clarified that he would rather it be a project for next year.
This is believed to be due to brand safety concerns, as celebrities want to ensure their image is not used to make problematic statements. Meta hopes the addition of personalities like D’Amelio and MrBeast will attract a younger audience that has shunned Facebook and migrated to TikTok, its main competitor. In addition to its smart assistants, Meta has also launched smart stickers. However, this feature was criticized as it was hijacked and users managed to generate wild images such as a child soldier and an obscene image of Justin Trudeau.
Additionally, Meta has been criticized for its willingness to pay creators and celebrities huge sums to use their images and voices, while working behind the scenes to make the systems less toxic. Because users are underpaid. AI companies like OpenAI and Meta are very critical of this practice. This year, a report revealed that OpenAI used outsourced Kenyan workers paid less than $2 an hour to scour some of the internet’s darkest corners and develop an AI filtering system to be integrated into ChatGPT.
According to the report, Kenyan workers earned between $1.32 and $2 per hour. According to some analysts, this amount is paltry for a company like OpenAI, which has raised hundreds of millions of dollars in recent years. Microsoft invested more than $1 billion in OpenAI in 2019 and announced earlier this year that it would invest several billion dollars more in the AI startup. Microsoft has not disclosed the size of the new investment, but some sources say it is $10 billion. Other investors are also expected to participate in this financing round.
Meta also criticized that it is willing to pay such an amount to celebrities/creators, but is reluctant to pay artists, researchers, writers and many others for the use of their works in the AI training data. This is how celebrities and designers can receive millions of dollars for their participation in the development of AI chatbots, but writers and scientists still struggle to make their voices heard, we read in the comments. Other concerns relate to the emotional connection these new AI assistants could create in users.
So far, efforts to enable talent to expand their personal brands have been marginal, with some celebrities or creators cloning themselves to create and market AI replicas as interactive experiences for fans. In May, a popular Snapchat creator launched CarynAI, an AI-powered girlfriend that offers fans personalized interactions for a dollar a minute. Celebrities also explored their own options, with some offering their perspective on AI-powered customer experiences in collaboration with brands. But the initiative was described as a fraud.
Meta’s integration of AI chatbots is also part of an emerging category of AI-based conversational applications that are positioning themselves to offer unique services or user experiences and types of interactions. OpenAI’s ChatGPT and Google’s Bard are positioned primarily as question answering services or productivity aids.
But individual chatbots have also emerged that offer users forms of entertainment and even companionship – be they living or dead famous people (e.g. Character.ai), therapists offering advice or emotional support, or romantic partners who Contact or simply offer “someone” to talk to (e.g. replica).
And you ?
What is your opinion on this topic?
What do you think about the amount Meta has paid to the celebrities behind its AI chatbots?
How do you assess the criticism of Meta in the context of this project?
Are these AI assistants promising? For what ?
What concerns do these AI assistants raise?
See also
Meta introduces Meta AI and 28 AIs with celebrities such as Snoop Dogg, Kendall Jenner and Paris Hilton to improve how young people interact with AI
OpenAI tapped Kenyan workers for less than $2 an hour to make ChatGPT less toxic. Experts say the rate is low because of employees’ “thankless tasks.”
Meta trained its AI using your public posts on Facebook and Instagram