The AI ​​is learning from what you said on Reddit, Stack Overflow, or Facebook. Are you okay with this?

CAMBRIDGE, Mass. (AP) — Post a comment on Reddit, answer coding questions on Stack Overflow, edit a Wikipedia entry or share a baby photo on your public Facebook or Instagram feed and you too will be helping train the next generation of artificial intelligence.

Not everyone agrees with this – especially as the same online forums where they spent years contributing are increasingly inundated with AI-generated comments that mimic what real humans might say.

Some old users tried to delete their previous contributions or rewrite them into something gibberish, but the protests didn’t have much effect. Several governments – including Brazil’s privacy regulator on Tuesday – have also tried to intervene.

“A more significant portion of the population feels helpless,” said Sarah Gilbert, a volunteer Reddit moderator who also studies online communities at Cornell University. “There’s nowhere to go except going completely offline or not contributing in a way that adds value to them and value to others.”

Platforms are responding – with mixed results. Check out Stack Overflow, the popular hub for computer programming tips. First, it banned responses written by ChatGPT due to frequent errors, but now it is partnering with AI chatbot developers and has punished some of its own users who tried to delete their previous contributions in protest.

It’s one of several social media platforms facing user caution — and occasional uprisings — as they try to adapt to the changes brought by generative AI.

Software developer Andy Rotering of Bloomington, Minnesota, has used Stack Overflow daily for 15 years and said he worried the company “could be inadvertently harming its greatest resource” — the community of contributors who have donated time to help other programmers.

“Keeping employees encouraged to provide feedback should be key,” he said.

Stack Overflow CEO Prashanth Chandrasekar said the company is trying to balance the growing demand for instant chatbot-generated coding assistance with the desire for a community “knowledge base” where people still want to post and “be recognized.” for their contributions.

“Fast forward five years – there will be all kinds of machine-generated content on the web,” he said in an interview. “There will be very few places where there is truly authentic and original human thought. And we are one of those places.”

Chandrasekar readily describes the challenges of Stack Overflow as one of the “case studies” he learned at Harvard Business School of how a company survives – or doesn’t – after disruptive technological change.

The story continues

For more than a decade, users typically came to Stack Overflow after typing a coding question into Google, then finding the answer, copying and pasting it. The answers they were likely to get came from volunteers who accumulated points to measure their credibility – which, in some cases, could help them get a job.

Now programmers can simply ask an AI chatbot – some of which are already trained on everything ever posted on Stack Overflow – and it can spit out an answer instantly.

ChatGPT’s debut in late 2022 threatened to drive Stack Overflow out of business. So Chandrasekar created a special team of 40 people at the company to accelerate the launch of his own specialized AI chatbot called Overflow AI. The company then made deals with Google and ChatGPT maker OpenAI, allowing AI developers to leverage the Stack Overflow question and answer archive to further improve their AI large language models.

This type of strategy makes sense, but it may have come too late, said Maria Roche, an assistant professor at Harvard Business School. “I’m surprised Stack Overflow wasn’t working on this sooner,” she said.

When some Stack Overflow users attempted to delete their previous comments following the Open AI partnership announcement, the company responded by suspending their accounts due to terms that make all contributions “perpetually and irrevocably licensed to Stack Overflow.”

“We quickly addressed the issue and said, ‘Look, this is not acceptable behavior,’” Chandrasekar said, describing the protesters as a small minority among the “few hundred” of the platform’s 100 million users.

Brazil’s national data protection authority took action on Tuesday to ban social media giant Meta Platforms from training its AI models on Brazilians’ Facebook and Instagram posts. It established a daily fine of 50 thousand reais (US$8,820) for non-compliance.

Meta, in a statement, called it “a setback for innovation” and said it has been more transparent than many industry peers who conduct similar AI training on public content, and that its practices comply with Brazilian laws. .

Meta has also encountered resistance in Europe, where it recently suspended its plans to start feeding people’s public jobs into training AI systems – which was supposed to start last week. In the US, where there is no national law protecting online privacy, this type of training is likely already happening.

“The vast majority of people simply have no idea that their data is being used,” Gilbert said.

Reddit has taken a different approach – partnering with AI developers like OpenAI and Google, while making clear that content cannot be mass-sourced without approval of the platform by commercial entities “without regard to the rights of the user or privacy.” Business helped bring Reddit the money it needed to debut on Wall Street in March, with investors raising the company’s value to close to $9 billion, seconds after it began trading on the New York Stock Exchange.

Reddit didn’t try to punish users who protested — nor could it easily do so, given the influence that volunteer moderators have over what happens in its specialized forums, known as subreddits. But what concerns Gilbert, who helps moderate the “AskHistorians” subreddit, is the growing stream of AI-generated comments that moderators must decide whether to allow or disallow.

“People come to Reddit because they want to talk to other people, they don’t want to talk to bots,” Gilbert said. “There are apps where they can talk to bots if they want. But historically Reddit has been used to connect with humans.”

She said it’s ironic that the AI-generated content threatening Reddit originated from the comments of millions of human Redditors, and “there’s a real risk that it could eventually end up getting people kicked out.”

——

Associated Press writer Eléonore Hughes in Rio de Janeiro contributed to this report.

——

The Associated Press and OpenAI a licensing and technology agreement which allows OpenAI access to part of the AP text files.