yego.me
💡 Stop wasting time. Read Youtube instead of watch. Download Chrome Extension

Meta's Creepy AI Celebrities


3m read
·Nov 4, 2024

Processing might take a few minutes. Refresh later.

What if you were able to have your loved ones live on with you long after they're gone, to hear their voice, experience their laugh, get their advice, and tell inside jokes that only the two of you know? If someone told you they could make that happen, would you take them up on that offer?

In 2017, John Mayer, the CEO of artificial intelligence company Forever Voices, did just that. He developed the bot version of his father who recently passed away. He could chat with his dad whenever he wanted, engage with him, and for a moment, escape the pain of him being gone. Since then, the AI market for bots based on real people, influencers, or celebrities has exploded. Companies have been built and rebuilt to capitalize on the AI craze, but none has more potential for influence than this one: Meta.

So when Meta introduced its new AI features, tech reporters and regular users alike leaned in. Meta's new features include customized stickers, IM editing, and an AI assistant. One development in particular that's thrown everyone for a loop is a new cast of AI bots. These bots aren't your run-of-the-mill AI bots, though; each one of them has a unique backstory and expertise in a particular niche. They have profiles on Instagram and Facebook, and most importantly, they're voiced by cultural icons and influencers like Tom Brady, Naomi Osaka, Kendall Jenner, Mr. Beast, and Paris Hilton.

But confusingly, the characters are different from their instantly recognizable celebrity voices. You're not chatting sports with Tom Brady, but rather a guy named Brew, who just so happens to look and sound exactly like Tom Brady. You can talk Dungeons and Dragons with the Dragon Master voiced by Snoop Dogg or look for advice from Kendall Jenner AI, your no BS ride or die companion. Some of these characters, like Jenner, make sense, while others leave you wondering what the connection is. For example, Paris Hilton is a crime-solving detective. What's the connection there?

Ironically, these bots were unveiled at Meta's annual product showcase, Connect, at the same time the actor's union, the Screen Actor Guild, was on strike harshly over demands around limiting AI-generated content that threatens to put actors out of work. So how did Meta get a bunch of non-actor celebrities to give away their likeness? Well, they didn't give it away at all. They were reportedly paid up to 5 million each for 6 hours of work and endless usage of their face and voice. Meta's deep pockets and cutting-edge AI technology called Llama positioned the company perfectly to take on such a high-profile AI project.

Unfortunately, a lot of the new bots are generally regarded as creepy and confusing. Chatting with AI Tom Brady or Brew might be a fun novelty at first, but quickly can evolve into a far less interesting conversation about football than one might expect with the actual Tom Brady. Novelty, it turns out, wears off pretty quickly.

So why is Meta taking such a big chance on this new Chad bot program that seems doomed to fail from day one? Well, just like many others, it's trying to win the artificial intelligence market. There's never been a more exciting time and competitive time for AI, and Meta is trying to do things a little differently than its main competitors like OpenAI. Llama, its homegrown tech, is open source, which means Meta is giving developers around the globe access to its software. This is in stark comparison to the technology behind ChatGPT, which OpenAI keeps under wraps.

Meta compares this strategy with Linux, an open-source PC alternative to Windows in the '90s and 2000s. Linux made its way into corporate servers worldwide and became a key component of the modern market. Meta is hoping that Llama will have the same effect in their eyes, by making the technology open source and allowing third parties to make improvements. This could result in better efficiency and ultimately make it cheaper for Meta to run the AI software.

And what better way to keep its software relevant than creating a pop culture moment using Snoop Dogg or Paris Hilton AI bots? Ultimately, the idea isn't that original; it's the same concept used by another company called Replica.

More Articles

View All
How Many 5 Year-Olds Could You Fight? -- And 18 Other DONGs!
Hey, Vsauce. Michael here. And I am now living in London. Besides popping on over to Disneyland Paris, I’ve also been looking at DONGs: Things you can Do Online Now, Guys. For instance, because I’m now in Britain, my team has changed for clickclickclick.…
Safari Live - Day 333 | National Geographic
This program features live coverage of an African safari and may include animal kills and carcasses. Viewer discretion is advised. This program features live coverage of an African safari. It’s a wonderful Sunday, and you are looking at a lovely drawn go…
What The Ultimate Study On Happiness Reveals
This video is about one of the most important questions: what leads to a happy life? Realistically, money. Being wealthy is definitely a big aspect of it. To save a lot of money. Money. Money. Earning money. It’s very important to be rich. It’s ea…
10 Low Cost Businesses To Start In A Developing Country
The best way to start making money in a developing country is to start a business for two reasons. One, there isn’t anything much to do anyway; and two, starting a business in that environment is way easier than anywhere else. That’s because all you have …
Developing an American identity, 1800-1848 | US history | Khan Academy
In this video, I want to take a look back at the period from 1800 to 1848, kind of from a bird’s eye view. This is a huge time in American history. In 1800, the United States was just a fledgling nation, less than 20 years out from winning its independenc…
Partial derivative of a parametric surface, part 2
Hello, hello again! So in the last video, I started talking about how you interpret the partial derivative of a parametric surface function, right? Of a function that has a two-variable input and a three-variable vector-valued output. We typically visual…