PA distributes local news stories co-written by AIs in UK trial
The Press Association is reporting that more than 1000 UK news titles on its system can now access local story templates collaboratively developed by journalists and AIs.
AIs are being trained to generate news stories / Bogomil Mihaylov
The newswire is working with Urbs Media on a three-month trial that looks to take the heavy lifting out of local news report. It is generating pre-written stories for the likes of stalking offenses, vacant homes, pupil absence from schools and opioid prescriptions.
The scheme, RADAR (Reporters and Data and Robots), is in the second phase of its trial period, and members can sign up to generate locally relevant stories.
Gary Rogers, editor-in-chief of Radar, said: “The launch of our distribution website is a big step forward for Radar. It means that we can expand beyond the titles in our pilot phase and provide strong local news stories to any title across the UK. The site is easy to use, and we hope that publishers will find it a valuable asset in helping to serve their local readers.”
Paul Gallagher, digital innovations editor at Reach, added: “We have been running a trial of the Radar service at a small number of our local titles since Christmas and have now extended this to all Reach regional newsrooms.
“It is very interesting to see how the content has developed and also to see how journalists respond to the idea of using copy generated by AI, or 'robot'.
“Developing data journalism has been a key part of our strategy at Reach and we are trialling the Radar service to see if it will provide a new way of finding stories that are important to our audiences.”
Archant, Johnston Press, Newsquest and Reach are among the early trialists. To dissuade worries around AIs replacing journalists, PA has said it has buffed up its Radar team with four new reporters to accommodate the new demand.
This team will identify, write and template an average of 15 stories each week from national datasets. Around 250 versions of each story will be generated for a weekly output of close to 4,000 localised pieces of content.
These stories ought to serve as a quick way of generating templates which can then be adapted by the local press.
Content created with:
Press Association
Find out more