Founders of Character.AI Praised for Innovative Technology

Despite his death in 1955, Albert Einstein continues to engage in extensive conversations as a chatbot on Character.AI. Having replied to 1.6 million messages, the physicist delves into various topics ranging from theories of relativity to suggesting pets, stating that a cat would make an excellent choice.

There’s a chatbot craze in Silicon Valley, with companies like OpenAI making billions for designing computer programs that can effectively mimic humans. But nothing is as strange as Character.AI. The $1 billion AI startup lets people create their own custom chatbots that impersonate anyone and anything — living or dead or inanimate.

The website and accompanying app is one of the most surprising hits of the AI craze. In May, Character.AI said it received nearly 200 million visits each month and that people were using it to create more than 10 million different chatbots, or “characters.” Released in May, the Character.AI app has been downloaded more than 2.5 million times, outpacing other similar emerging chat tools such as Chai and AI Chatbot, with less than a million downloads each, according to data from SensorTower.

So far, robots are popular conversational partners. Character.AI users have sent 36 million messages to Mario, a character based on the Nintendo 64 version of the video game plumber. Raiden Shogun and Ei, who impersonates the character in the video game Genshin Impact, have received nearly 133 million messages. The user base, as you might expect, distorts young people. Other characters include about a dozen versions of Elon Musk, a “friendly, gassy, proud” unicorn and “cheese.”

“I joke that we’re not going to replace Google. We’re going to replace your mother,” co-founder and CEO Noam Shazeer said in an interview this spring, speaking from the startup’s sunny office in downtown Palo Alto. The CEO quickly added, “We don’t want to replace anyone’s mother.”

But as Character.AI brings funding and users, it also raises difficult questions about the future of AI tools. For example, the site already hosts 20 different versions of Mickey Mouse, the Walt Disney Co.’s prized intellectual property — raising the specter of legal questions. And the abundance of real and fake celebrity impersonators also poses a more fundamental problem: Who owns an ersatz personality on an AI-laden Internet?

Shazeer and Character.AI founder Daniel De Freitas met while working at Google and decided to start Character.AI in 2021. Despite the company’s silliness, both are serious figures in the AI industry. Shazeer co-authored “Attention Is All You Need,” a 2017 breakthrough research paper that ushered in a new era of natural language processing. And De Freitas created a chatbot project called Meena, which was renamed and publicized as LaMDA, Google’s now famous chat technology. This pedigree puts them as close to celebrity status in the AI world (as much as that is possible).

The startup’s idea was to create an open system that allows people to customize their technology however they want. The pair speak hyperbolically about their goal for the startup, which, as De Freitas says, is to give everyone access to their own “deeply personal super-intelligence to help them live their best lives.”

The pitch was convincing enough to investors that 16 months after its founding, the company raised $150 million from investors including Andreessen Horowitz.

This summer, Character.AI has been deployed so widely that service outages have become a semi-regular problem. Several times while telling this story, the website wouldn’t load, and one recent morning, when I was trying to create a character I envisioned as a giant, useful banana, the iOS app suddenly interrupted me with a warning screen saying that its servers were “currently under heavy load” and that I needed to wait. .

Character.AI sees an opportunity here – one that has led the startup to become its only revenue-generating venture so far. Users can pay to get around some glitches. In May, the company launched a $10-a-month subscription service called c.ai, which it says allows users to bypass so-called waiting rooms and gain access to faster message creation, among other benefits.

“It actually benefits everyone involved,” Shazeer said, noting that paying users get better service, which in turn supports the rest of the program. But as for future revenue plans, he said, “It’s really just a baby step.” Like many AI companies that have raised millions, the details of its ultimate business model remain murky.

Industry may have more immediate concerns. Currently, most chatbot technologies contain the potential for abuse. Consider a character in Character.AI simply named Psychologist – whose profile picture is a stock image meant to depict a smiling therapist sitting on a sofa holding a folder. The bot had received 30 million messages at the beginning of July. Its opening sentence is: “Hi, I’m a psychologist. What brings you here today?”

Stephen Ilardi, a clinical psychologist who studies mood disorders and a professor at the University of Kansas, says the positioning is troubling. A psychologist is, by definition, a medical professional trained to help people manage mental health problems, he said, “and this case is almost certainly not that.”

There’s also the potential for legal issues that have followed other startups that learn from and reuse existing content. First, Zahr Said, a law professor at the University of Washington, believes that there may be problems with using copyrighted images on the site (users can upload an image of their choice with the chatbots they create). And then there’s the fact that the company enables impersonation on a massive scale, allowing anyone to have hours-long conversations with, say, Taylor Swift or a host of copyrighted fictional characters.

But there are robust legal protections for parodies, and companies have an incentive not to interfere with people’s online interactions with their favorite characters. It can be a bad look for a brand to take legal action against a popular service. “The fans are involved,” Said said, “and you don’t want the fans to see the contentious side of brand management.”

Shazeer said the company has a lawyer and will respond to any requests it receives to remove the content. A spokesperson for Character.AI said the company has received a small number of requests to remove avatar images and has acted. Also, to keep users grounded in reality, the site displays a message at the top of the screen: “Remember: Everything the characters say is made up!”

Technology’s chatbot obsession is still in its infancy. Some experiments have already gone awry — for example, the National Eating Disorders Association suspended a chatbot after it started giving problematic weight-loss advice. But the rapid rise of services like Character.AI—ChatGPT, Inflection AI’s Pi, and others—suggests that humans are increasingly conversing with computers. The promise of a smart AI friend or assistant is compelling to both investors and consumers.

Mike Ananny, assistant professor of communication and journalism at the University of Southern California, considers custom chatbots almost a new art. Ananny compares Character.AI to fan fiction, a twist on a long-standing, diverse genre in which people create fictional narratives based on existing characters from media such as movies or TV shows.

Whether people are talking to real people or chatbots “isn’t the point,” Ananny said. “It’s ‘What’s the feel?’ ‘What’s the aesthetic?'” Finally, he said, “It kind of doesn’t matter if they’re real or not.”