testing journalism AI – NYT

36 testing journalism AI – NYT

The tech large’s AI chatbot Bard is already infamous for serving up false info as factual

Google is testing an AI-powered journalism product and pitching it to main information organizations, the New York Occasions reported on Thursday, citing three sources near the matter. The Occasions was allegedly one of many shops approached by Google.

Identified internally as Genesis, the device is able to producing information tales based mostly on person inputs – particulars of present occasions like who, what, the place, or when, the sources stated. The corporate allegedly sees it as “accountable expertise” – a middle-ground for information organizations not serious about changing their human workers with generative AI. 

Along with the creep issue – two executives who noticed Google’s pitch reportedly referred to as it “unsettling” – Genesis’ mechanized method to storytelling rubbed some journalists the incorrect means. Two insiders informed the Occasions it appeared to take as a right the expertise required to supply information tales that weren’t solely correct however well-written.

A spokeswoman for Google insisted Genesis was “not meant to…substitute the important function journalists have in reporting, creating, and fact-checking their articles” however may as an alternative supply up choices for headlines and different writing kinds. 

One supply stated Google truly seen Genesis as extra of a “private assistant for journalists,” able to automating rote duties in order that the author may give attention to extra demanding duties, like interviewing topics and reporting within the discipline.

The invention that Google was engaged on a “ChatGPT for journalism” sparked widespread concern that Genesis may open a Pandora’s Field of faux information. Google’s AI chatbot Bard shortly turned notorious for spinning up advanced falsehoods and providing them as reality following its introduction earlier this yr, and CEO Sundar Pichai has admitted that whereas these “hallucinations” look like endemic amongst AI massive language fashions, nobody is aware of what causes them or hold an AI trustworthy. 

Worse, Genesis may marginalize actual information if Google encourages its adoption by tweaking its search algorithms to prioritize AI-generated content material, radio editor Gabe Rosenberg tweeted in response to the New York Occasions’ article. 

A number of well-known information shops have dabbled with utilizing AI within the newsroom, with lower than inspiring outcomes. BuzzFeed went from utilizing AI to generate personalized quizzes to churning out dozens of formulaic journey items to asserting all content material can be AI-generated in beneath six months, regardless of having promised its writers again in January that their jobs had been secure. 

CNET was caught earlier this yr passing off AI-written articles as human content material and utilizing AI to rewrite outdated articles in an effort to artificially improve their search engine rankings. 

Regardless of these disasters, OpenAI, the corporate accountable for ChatGPT, not too long ago started signing offers with main information organizations just like the Related Press to encourage the expertise’s adoption within the newsroom.

You may share this story on social media:

Supply hyperlink