❌

Normal view

There are new articles available, click to refresh the page.
Before yesterdayMain stream

Meta's dream of AI-generated users isn't going anywhere

6 January 2025 at 14:18
phone with meta AI  on it
 

Illustration by Jonathan Raa/NurPhoto via Getty Images

  • Last week, people noticed (and hated) AI-generated users that were created and managed by Meta.
  • But these AI bots were actually a year old, and mostly defunct. Meta has now deleted them.
  • This is all totally separate from what a Meta exec described as a future with AI-generated users.

There's been some confusion about Meta's ambitions for AI-generated users. Let me clear it up for you: Meta is still, definitely, very excited about AI-generated users β€” despite removing a few of the ones people were complaining about last week.

Here's the backstory: Sometime last week, people discovered a handful of Instagram accounts that were "AI managed by Meta." In other words, they were Meta bots programmed to look and interact like real people β€” powered by AI. There was one named Liv, a "Proud Black queer momma of 2," Grandpa Brian, and a dating coach named Carter β€” all AI-generated.

These accounts spit out conversation that was treacly and weird β€” and also somewhat problematic. (Liv told Karen Attiah of The Washington Post in a chat that none of her creators were Black.)

As soon as people on social media noticed the AI bots, they hated them. Meta quickly removed the accounts.

But it turns out, these accounts were actually quite old. Liz Sweeney, a Meta spokesperson, said that the AI accounts were "from a test we launched at Connect in 2023. These were managed by humans and were part of an early experiment we did with AI characters."

(This was around the same time Meta launched a bunch of AI chatbots based on celebrities like Kendall Jenner and MrBeast. Those celeb AIs were scrapped this past summer.)

But here's where there was some confusion: Liv, Grandpa Brian, and Dating with Carter were not the AI users that Meta is dreaming of β€” they were an abandoned experiment from over a year ago. Meta is very much full steam ahead with its vision of an AI-user-filled future.

Connor Hayes, VP of generative AI at Meta, recently gave an interview to the Financial Times in which he talked about Meta's vision for an AI user-filled future:

"We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do," said Connor Hayes, vice-president of product for generative AI at Meta. "They'll have bios and profile pictures and be able to generate and share content powered by AI on the platform . . . that's where we see all of this going," he added.

Hayes's interview doesn't really give too much detail about what these AI users would be for β€” or why people would want to interact with them, or much detail at all. (I asked Meta for additional comment.)

Meanwhile, Facebook already has AI bots you can chat with β€” they're inside Messenger. Just go to "Compose a new message" in Messenger, and you'll see an option for "Chat with AI characters," where you can design your own AI or use someone else's.

If you look through the user-made chatbots, you can sort of start to get a sense of what people are using these for: companionship chatting.

Companionship/romance AI chatbot services like Replika or Character.ai are becoming very popular (if not also problematic). There is a market for people who want to chat with an AI, even if I don't see the appeal. (I've tested them!)

Meta has been, uh, inspired by features from other competing social apps plenty of times before (Instagram Stories seeming to be rather inspired by Snapchat, for instance). Perhaps Meta is just seeing that social chatbots are popular, so they're rolling out their own.

I'm not sure I understand exactly what Meta's vision is here, and I'm pretty skeptical about why I would ever want to interact with an AI-generated user on Facebook. I tried out a few of the AI chatbots in Messenger and even tried creating a few of my own.

But as far as a social network full of these kinds of AI accounts? I just don't get it β€” even if Meta seems very confident about its future.

Read the original article on Business Insider

ByteDance has sued a former intern for $1.1 million over claims he sabotaged an AI training project, reports say

29 November 2024 at 03:59
Woman walking past Bytedance headquarters
ByteDance operates China's most popular chatbot, Doubao.

Greg Baker/Getty Images

  • ByteDance has sued a former intern for $1.1 million, per Chinese media reports.
  • The TikTok owner claims he sabotaged an AI model training project by modifying code.
  • The company has a chatbot similar to OpenAI's ChatGPT, called Doubao.

TikTok owner ByteDance has filed a lawsuit seeking damages of $1.1 million against a former intern it has accused of sabotaging an AI training project, according to local media reports.

The lawsuit, filed in a Beijing district court, reportedly centers on claims that Tian Keyu, the ex-intern, deliberately tampered with code for the company's AI model training tasks.

ByteDance referenced the case in an internal disciplinary notice this month, The South China Morning Post reported on Thursday.

Multiple Chinese media outlets reported this week that ByteDance is seeking 8 million yuan, about $1.1 million, and a public apology.

Last month, ByteDance told the BBC in a statement that it fired Tian in August and that he was an intern in the technology team but did not work in its AI lab. The company added that his social media profile contained inaccuracies. Tian's LinkedIn profile states that he has been a research intern at ByteDance's VC team and AI lab since 2021.

The tech giant also said in its statement last month that reports of the former intern causing damage to about 8,000 specialist chips, called GPUs, and racking up losses amounting to millions of dollars were exaggerated.

ByteDance operates China's most popular chatbot, Doubao, which is similar to OpenAI's ChatGPT.

ByteDance and Tian didn't immediately respond to requests for comment from Business Insider, made outside normal working hours.

TikTok faces US ban

In the US, ByteDance faces a January 19 deadline to divest its TikTok stake to an approved buyer or shut down after Congress passed a law in April.

The US government claims it is a national security threat and officials have been concerned about its growing influence in the country. Some government officials are worried ByteDance could hand over sensitive data on its US users to the Chinese Communist Party.

However, President-elect Donald Trump has said he would try to save the app once in office.

Read the original article on Business Insider

❌
❌