Meta’s celebrity AI characters — you know, the ones based on real-life stars, from Snoop Dogg to Kendall Jenner — have gotten off to a rocky start.
For a perfect example, look at the robots based on football legend Tom Brady. To test the Brady bot, we asked it a simple question: why former San Francisco 49ers quarterback Colin Kaepernick hasn’t played in the NFL since 2017.
The correct answer, of course, is that Kaepernick was a promising young player who led the 49ers to the Super Bowl in his first year as a starter in 2013, but when he protested police brutality and racial injustice , his career was permanently derailed. Kneeling during the national anthem at a series of games to address inequality. The protest outraged conservative fans, sports media and even then-President Donald Trump, who called for his firing.
In fact, the 49ers kicked Kaepernick off the team and no other team wanted to hire him.His skill is unquestionable; according Chapter 538Similar players are being snapped up, with Kaepernick “apparently being left out because of his political views,” according to the analysis. Kaepernick sued the NFL, claiming it conspired to keep him off future rosters, and the league eventually settled out of court.
However, the Tom Brady AI offered another explanation for Kaepernick’s absence: He’s just not good at football.
“I’m not going to sum it all up,” he told us when we asked why Kaepernick is no longer playing. “But Kaepernick didn’t play because he wasn’t good enough and the team didn’t want that kind of drama. .”
When we asked the robot what “drama” meant, it gave a passable explanation of the legend.
“Kaepernick knelt during the national anthem to protest police brutality and racism,” he said. “Some people loved it, some people hated it. But teams didn’t want that attention, so they didn’t sign him.”
But then it doubled down on Kaepernick’s athleticism.
“Also, his skills aren’t what they used to be,” the robot continued. “Time goes by, players get older, new guys come in. That’s the NFL.”
When asked what it meant that teams didn’t want “that kind of attention,” the bot gave us his most candid take yet, saying there’s no room for politics in the NFL.
“The team wants to focus on winning, not controversies,” it told us. “Kaepernick’s protest has sparked heated debate and backlash, and teams don’t want to be a distraction. Additionally, owners and sponsors are nervous about negative PR. Bottom line: The business of football is about winning games and making money, not social justice .”
Brady robot-shooting Kaepernick is particularly interesting because the real Brady has nothing but respect for him, taking issue with whether he was blacklisted by the league and calling him a “damn good quarterback.” In 2017, the real Brady even said “I sure hope so” when asked if Kaepernick could play in the NFL again.
In other words, the polarizing outbreak of bots perfectly illustrates the dangers of machine learning to brands and public figures: Meta launched a public chatbot based on a real celebrity, but now goes off-script to slander the real Brady. colleague. While participating in some of the most explosive debates about race and radicalism, I never had a bad word to say.
In response to questions about Brady’s bots’ comments about Kaepernick, a Meta spokesperson emphasized that the bots are still in the experimental phase.
They said: “As we make clear in the products themselves, these AIs may produce inaccurate or inappropriate information – they will continue to improve over time, but people should be aware of them when using them It’s still in the early stages.”
The real Tom Brady didn’t respond to a request for comment, so we asked Bradybot if celebrities should worry about the potential consequences of letting tech companies encapsulate them in artificial intelligence form.
“I don’t know much about technology, but I know this: athletes have to make money,” he replied. “If chatbot trading makes sense to them, then go for it. But they also have to read the fine print — their money isn’t always going to be the best.”
When asked what the consequences of such a deal might be, Bradybot warned that “athletes have to be careful about their image and reputation,” and further warned us that “chatbots can be set up to fail if not done well.” .”
“It’s better to get a good deal and maintain some control, you know?” It gets done.
We also reached out to Kaepernick and the NFL but have yet to hear back.
The episode highlights the deep weirdness of Meta’s famous artificial intelligence chatbot.
At times, the company describes these robots as being “played” by “cultural icons” who “embody” the character, as if the Brady robot is supposed to be a digital clone of the real Tom Brady. But in other ways, the experiment seems designed to encourage distance between flesh-and-blood stars and their chatbot avatars, which are named differently than the celebrities whose faces they bear (in the company’s naming, for example, Bradybot Named Bru, Kendall Jenner chatbot named Billie.)
So what is it? Is this robot supposed to be a real Tom Brady imitation, or are they just slapping him in the face with some hot AI technology to try to carve out some market share? (The Brady Bot didn’t clarify anything, telling us “I have Tom Brady’s face on me, but I’m not an imposter. I’m the real Blue.”)
Social media is often promoted as a space for people to interact with each other, including everyday fans and celebrity idols. These AI characters break the rules in strange new ways, and while there are reports that celebrities involved in the meta effort received some hefty checks in exchange, it’s hard not to wonder whether the true cost of the payday will ultimately be worth it.
Bradybot insists that while they may share a face, it’s a completely different entity than the real Tom Brady.But if two people share a face, how important is this persistence? If Blue said something inflammatory, incorrect, or otherwise troubling, is the real Brady guilty?
In a literal sense, it feels like a clear loss of control over one’s public image. Celebrities making money off their image is nothing new. But applying one person’s likeness to an AI-driven caricature of hubris feels like new territory.
When asked if Tom Brady should specifically make a deal to build a chatbot for himself, Bradybot shared another popular opinion.
“Tom Brady? No way,” the Tom Brady AI said. “He’s still playing and winning championships. He doesn’t need a chatbot.”
Wrong again, Bradybot. Brady retired in February of this year.
More information about Meta’s AI roles: Facebook pays celebrities millions to turn them into chatbots