Ask Microsoft Corp.’s new Bing chatbot about Donald Trump’s indictment and the artificial intelligence tool delivers a response that strikes fear in media executives.
(Bloomberg) — Ask Microsoft Corp.’s new Bing chatbot about Donald Trump’s indictment and the artificial intelligence tool delivers a response that strikes fear in media executives.
The three-sentence summary seems helpful, providing links to news outlets like CNN and the Washington Post. It’s also friendly: “Is there anything else you would like me to help you with?”
But media executives see the technology, commonly known as generative AI, as a new existential threat. They worry people will find chatbot summaries of articles good enough and not visit their websites, stealing readers and advertisers, as earlier internet innovations did. And they’re looking at ways to fight back, whether it’s through walled-off content, legislation or payment for their work.
Some publishers have discussed trying to block AI chatbots from ingesting their articles. By editing a file called robots.txt, they could try to keep bots from accessing their websites.
It’s unclear that will work. If AI chatbots collect information the same way as search engines, “publishers ought to have the ability to control which parts of their content are visible for possible inclusion, while marking other content as restricted,” said Francesco Marconi, a computational journalist and co-founder of the AI company AppliedXL who has worked for the Wall Street Journal and Associated Press.
“However,” Marconi added, “the actual operation of these AI systems remains unclear.”
Some media executives say there’s nothing they can do to protect their content, because new chatbots gather information from search results, and publishers don’t dare exclude their articles from search engines because of the traffic they provide. Above all, they don’t know enough about how the rapidly evolving technology works.
“What happens inside the machine is not transparent,” said Dietrich von Klaeden, senior vice president of public affairs at the German media giant Axel Springer SE, which owns online publications like Politico and Insider.
Artificial-intelligence chatbots have captivated the business world since the research lab OpenAI introduced ChatGPT in November. In recent weeks, Microsoft introduced an AI-powered chatbot that uses technology from OpenAI, and Alphabet Inc.’s Google released a chatbot called Bard.
Unlike typical search results, which reveal a snippet of an article and a link to a website, AI services can produce paragraph-long responses.
Google said in a statement it’s using Bard as “an opportunity to learn and get feedback from a range of stakeholders, including news publishers” and would “continue to prioritize approaches that will allow us to send valuable traffic to news publishers and support a healthy, open web.”
A spokesperson for Microsoft said the company intends to collaborate with news outlets. “We are constantly working with publishers,” the person said. “And have great respect for the content they create.”
OpenAI didn’t respond to requests for comment.
News executives are trying to assess the potential impact on their businesses. Digital Content Next, an industry group whose members include the New York Times, News Corp. and Bloomberg, has held more than 10 meetings in the past three weeks about the implications of AI chatbots. Some executives say it could force them to rethink their strategies for attracting readers and rely more on other methods, like newsletters.
While publishers see potential threats, they also see opportunities. They envision ways that AI can cut costs and make their newsrooms more productive.
“It can make reporters’ work more efficient, enabling them to concentrate on journalistic creation, on investigative work and on relevant commentary,” von Klaeden said.
Media executives are also pushing for changes to copyright law, saying companies like Microsoft and Google should be required to prove they have the right to use articles for AI.
One unsettled legal question is whether AI chatbots are engaging in “fair use” under copyright law.
“If you are taking someone’s content and using it to train something else in a way that reduces the value of content, that doesn’t feel like fair use to me,” said Dan Check, chief executive officer of Slate magazine.
Many publishers also want a cut of the advertising revenue that some AI chatbots generate using summaries of their journalism. At an investor conference in March, News Corp. Chief Executive Officer Robert Thomson said his company has started discussions about receiving payment from AI companies.
“Clearly they are using proprietary content, and there should be, obviously, some compensation for that,” Thomson said.
Bloomberg and the New York Times declined to comment.
Newsrooms face more than financial threats from AI. Chatbots have also proved adept at spreading misinformation, adding to the challenge that journalists face in earning public trust.
Publishers have a long and complicated history with search engines. For years, newspapers complained that digital media companies published posts based on their reporting with a Google-friendly headline and appeared higher in search results, stealing their readers. Today, many newsrooms train journalists on how to write search-friendly headlines because it remains an important source of traffic.
Now, news executives are trying to get ahead of another evolution in how readers stay informed.
“There is a new audience segment that is consuming more content than any other,” Marconi said. “And that is not composed of humans, but machines.”
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.