Google launches AI tool to help journalists write news articles

The paper reported that Google has held meetings with the NYT itself as well as the Washington Post and News Corp which owns the Wall Street Journal.

The search giant reportedly said the tool, which has been dubbed Genesis, will help journalists to produce different writing styles and offer options for headlines rather than pitching it as a full-scale replacement.

"These tools are not intended to, and cannot, replace the essential role journalists have in reporting, creating and fact-checking their articles," Google said.

But according to the NYT, some of the executives who were in the room while Google was pitching found the proposal "unsettling".

Earlier this month, the Associated Press signed a deal with OpenAI to explore the possibility of using generative AI to create news articles.

Kelly McBride, an expert in journalism ethics for the Poynter Institute, said use of such tools would "erode the quality" of news articles as well as confidence in news media.

"This is a whole new level of threat and it's not like we can turn back," she said.

The introduction of Genesis follows the launch of ChatGPT last year, a large-language model (LLM) which can answer questions in a seemingly natural way and is trained on a massive data set.

Education establishments have already faced difficulties with students using the tool to complete assignments. It has been shown to be able to create passing-grade answers at university level - including passing law exams at one university.

Google quickly responded with its own AI service, called Bard, that had similar functionality to ChatGPT.

However, at its inaugural unveiling in February, the chatbot faced heavy criticism after it was shown to provide inaccurate information in various promotional materials, causing the price of Google stock to plummet.

Bard is powered by a lightweight and optimised version of LaMDA, a conversational tool first introduced by Google in 2021. Bard will be updated with newer, more capable models over time that will be further enhanced through interactions with users.