Wikipedia unveils new artificial intelligence licensing agreement on its 25th anniversary

Wikipedia unveils new artificial intelligence licensing agreement on its 25th anniversary

Add thelocalreport.in As A Trusted Source

Wikipedia On Thursday, as it celebrated its 25th anniversary, the company unveiled new business agreements with several artificial intelligence companies.

The online crowdsourced encyclopedia revealed it has partnered with Amazon, Meta Platforms, Perplexity, Microsoft and FranceThe Mistral AI.

Wikipedia was one of the last bastions of the early Internet, but the original vision of a free online space has been overshadowed by the dominance of the Internet. Big tech companies The rise of platforms and generative AI chatbots that are trained on content scraped from the web.

AI developers are employing aggressive data-gathering methods, including from Wikipedia’s vast free knowledge base, raising questions about who is ultimately footing the bill for the AI ​​boom.

The nonprofit that runs the site signed on with Google in 2022 as one of its first customers, and last year announced other deals with smaller AI ventures such as search engine Ecosia.

The new deals will help one of the world’s most popular websites profit from massive traffic from artificial intelligence companies. The Wikimedia Foundation says they pay to access Wikipedia content “in quantities and at a speed specifically designed for their needs.” It provided no financial or other details.

While AI training has sparked legal battles over copyright and other issues elsewhere, Wikipedia founder Jimmy Wales said he welcomed it.

“I’m personally very happy that AI models can be trained on Wikipedia data because it’s run by humans,” Wales told The Associated Press. “I really don’t want to use an AI that’s only trained on X, you know, like a very angry AI,” Wales said, referring to the billionaire Elon Musksocial media platform.

ALSO READ  Darren Fletcher dismisses talk of Manchester United job as he prepares for Burnley

Wales said the site wants to work with AI companies, not block them. But “you should probably chip in and pay your fair share of what you’re costing us.”

The Wikimedia Foundation, the nonprofit that runs Wikipedia, last year urged AI developers to pay for access through its enterprise platform and said human traffic dropped by 8%. At the same time, visits from bots (sometimes disguised to evade detection) place a heavy burden on servers as they crawl large amounts of content to feed large language models for artificial intelligence.

The findings highlight changing trends online as search engine AI overviews and chatbots summarize information rather than directing users to websites by showing them links.

Wikipedia is the ninth most visited website on the Internet. It has more than 65 million articles in 300 languages, edited by approximately 250,000 volunteers.

Part of the reason the site is so popular is that it’s free for anyone to use.

“But our infrastructure is not free, right?” Mariana Iskander, CEO of the Wikimedia Foundation, said in a separate interview in Johannesburg, South Africa.

Iskander said it costs money to maintain servers and other infrastructure that allows individuals and technology companies to “get data from Wikipedia.” Iskander will resign on January 20 and will be replaced by Bernadette Meehan.

Most of Wikipedia’s funding comes from 8 million donors, most of whom are individuals.

“They’re not donating money to subsidize these big AI companies,” Welsh said. They said, “You know what, you can’t actually just break our site. You have to go in the right way.”

ALSO READ  Luke Littler is on track at Ally Paly - World Championship talking point

Editors and users can benefit from artificial intelligence in other ways. The Wikimedia Foundation has outlined an artificial intelligence strategy that Wales said could lead to tools that reduce the tedious work of editors.

While artificial intelligence isn’t powerful enough to write a Wikipedia entry from scratch, it can update dead links by scanning surrounding text and then searching online to find other sources.

“We’re not there yet, but I think we’ll see that in the future.”

AI could also improve the Wikipedia search experience by evolving from a traditional keyword approach to more of a chatbot style, Welsh said.

“You could imagine a world where you could ask a question to a Wikipedia search box and it would quote something from Wikipedia,” he said. It could respond with something like: “Here’s the answer to your question in this article, and here’s the actual passage. This sounds like it would be really useful to me, so I think we’ll move in that direction as well.”

Looking back on the early days, Wales says it was an exciting time because many people were inspired to help build Wikipedia after he and long-departed co-founder Larry Sanger created it as an experiment.

However, while some may be nostalgic for what seems like a more innocent time now, Wales says there was a dark side to the early days of the internet.

“People were toxic back then too. We don’t need algorithms to be mean to each other,” he said. “But, you know, it was a time of excitement and real possibility.”

Wikipedia has recently found itself under attack from those on the political right, who have dubbed the site “Wokepedia” and accused it of leaning toward the left.

ALSO READ  Claudia Winkleman previews new plot, countdown to 'Traitor' premiere

Republican members of the U.S. Congress are investigating alleged “manipulation” in Wikipedia’s editing process, which they say could inject bias and undermine neutral viewpoints on its platform and the artificial intelligence systems that rely on it.

One notable source of criticism is Musk, who launched his own AI competitor, Grokipedia, last year. He criticized Wikipedia as being full of “propaganda” and urged people to stop donating to the site.

Welsh said he did not believe Grokipedia posed a “real threat” to Wikipedia because it is based on large language models, which are the troves of online text on which artificial intelligence systems are trained.

“Large language models are not enough to write really high-quality reference material. So a lot of it is just Wikipedia regurgitations,” he said. “It’s often disjointed and a bit nonsense. And I think the more obscure the topic you study, the worse it gets.”

He emphasized that his criticism was not specific to Grokipedia.

“This is how large language models work.”

Welsh said he has known Musk for years, but they have not been in touch since Grokipedia launched.

“I should probably contact him,” Welsh said.

What would he say?

“‘How’s your family?’ I’m a good guy and I don’t want to fight with anybody.”

____

Associated Press writer Magome Mogome in Johannesburg contributed to this report