Google is testing an artificial intelligence tool that can write news articles, in the latest evidence that the technology has the potential to transform white-collar jobs.

The product, known as Genesis, uses AI technology to absorb information such as details of current events and then create news stories. The tool has been featured in The New York Times, The Washington Post and Wall Street Journal owner News Corp as a “helper”, according to the New York Times.

Google said it was in the early stages of exploring the AI ​​tool, which it said could help journalists with options for headlines or different writing styles. He stressed that technology was not meant to replace journalists.

He said: “These tools are not intended to replace the role of journalists in reporting, creating and fact-checking their stories. Our goal is to give journalists the choice to use these emerging technologies in a way that improves their work and productivity, just as we provide users with support tools in Gmail and Google Docs.

Two New York Times executives who saw the pitch said it “seemed to take for granted efforts to produce accurate and astute reporting.”

The newspaper quotes a person familiar with the product who said the tool would serve as a “personal assistant for journalists” to automate certain tasks, and that Google saw it as an opportunity to help “steer the publishing industry away from the pitfalls of generative AI”.

The move comes after OpenAI and The Associated Press reached an agreement for the maker of ChatGPT to use the news agency’s story archive to train its AI models, which ingest large amounts of material to produce plausible answers.

In a report last month, accounting group KPMG estimated that 43% of the tasks performed by authors, editors and translators could be performed by AI tools. The Organization for Economic Co-operation and Development reported last week that major economies were on the cusp of an “AI revolution” that could lead to job losses in skilled professions such as law, medicine and finance.

Apple is testing an AI-powered chatbot that engineers call Apple GPT, according to Bloomberg. The chatbot is believed to have the potential to challenge ChatGPT, but Apple hasn’t released a clear plan on releasing the technology to consumers. The chatbot works from an AI model called Ajax.

Apple Chief Executive Tim Cook recently said the company is “looking closely” at AI technology.

Peter Welinder, an OpenAI executive, denied claims that the model underlying the most advanced version of ChatGPT, a model called GPT-4, was getting “dumber”.

Welinder said in a tweet that people are becoming such heavy users of it that they’re noticing issues they’ve never seen before. “We didn’t make GPT-4 dumber. On the contrary: we make each new version smarter than the previous one,” he said.

Users can check how responses have changed by entering a previously used prompt from the early days of ChatGPT and comparing the results. An user tweeted: “it generates faster, but the quality seems worse.”

Another writing on the OpenAI Developer Forum: “It’s totally awful now…It’s brain dead compared to before…if you didn’t really push it with what it could do before, you wouldn’t notice.” Still, if you really use it to the fullest, you see it’s obviously a lot dumber.

As newsrooms explore the possibility of using AI, a investigated this year by anti-disinformation firm NewsGuard discovered that bots were already powering dozens of AI-generated content farms.