DITA 22 Finalist Button_1080x1080 px9
Regional Finalist - no background


Ai in the Newsroom – 95% accurate is failure. 

In recent years, newsrooms have seen declining numbers of full and part-time staf, driven by declining advertising revenue. Established newspapers have not yet quite figured out how to monetise their online offerings, while their print versions continue to bleed dollars.  

Budgetary Pressures 

The budgetary pressures must create a temptation to cut staffing numbers and allow generative AI to create articles. This is a leap too far, too fast, too potentially damaging to a newspaper’s reputation. There are, however, time-saving ways in which generative AI can help journalists produce copy which is reliable, informative and generated by humans. 

Useful Gen AI Tools 

Gen AI has the obvious capacity to absorb very large tracts of text and extract the key elements from it. This is extremely useful for court judgements, long speeches in parliament or political manifestos, for example. The first step, in highlighting key points, can then drive a journalist to find those key points in the text, expand his/her knowledge around them and seek informed comment to enhance their article.  

At an earlier stage, AI can be used to filter Press Releases, abstract key pieces of information, and put them into diaries (date, venue, topic, key speaker etc.) for editors to review quickly. 

The art of shorthand is almost a distant memory now, and as useful to a journalist as speaking Latin!  Audio recorders take most of the information, enhanced with a few salient points and notes. AI can transcribe that audio into text, again saving a lot of time and money.  


But AI is not really ’intelligence’. It is language. AI doesn’t understand empathy, or accuracy, or the news lead for that matter! It needs a human to do that. 

In his presentation to the recent Notified webinar, Troy Thibodeaux from AP America (Associated Press) said that many of the AI salespeople will ‘brag’ about the fact that their tool is 95% accurate. Wow!. However, 95% accuracy is failure for journalists. Speaking at the webinar Troy said “If 5% of stories have correctives, we’re out of business” 

Finally, there are ethical issues for journalists using AI. The source of news should be transparent, so the use of AI should be clearly identified. And, it goes without saying, that everything should be proofread and signed-off by a human before publication.  

Connect with the author on LinkedIn.  

Webinar organized by Notified, held 01 November 2023. Two participants: Andrea Baillie, The Canadian Press and Troy Thibodeaux, AP (Associated Press) USA 

Harnessing the power of Technology - Creating the Thought Leaders of Tomorrow