AI in the newsroom and how the journalist role has changed
Opinion – Week in Focus
Ray Snoddy joins our AI-focused editorial special by interviewing both the FT‘s Chris Cook and ChatGPT on the growing influence of AI on newsrooms.
For freelance journalists, AI is the best thing since sliced bread, although you must not swallow it whole and always remember to enquire where the ingredients come from.
The best thing of all for journalists toiling alone in modest home offices or bedrooms without secretaries, assistants or research teams is that it almost instantly puts you on a par with the big battalions.
Sometimes, in a matter of seconds, it is possible to bring together a mass of information that in olden times would have taken hours, more likely days, to bring together, if it were possible at all.
It really is the great equaliser – individual journalists absolutely have more than a chance of competing with team players in large media organisations.
It’s almost like magic. But is it too good to be true?
AI only works if you are careful and understand its limitations. It can only pull up information that already exists and is publicly available, and that can, of course, bias the outcome.
Rigorous checks on sources are necessary, and the ancient computer axiom- rubbish in, rubbish out-still applies in spades despite the apparent sophistication of the latest tools.
It can only deal with the past and is hopeless at predicting the future; if you have any sense, avoid AI conclusions or treat them with extreme caution.
There is also one small problem for freelancers hoping to make a living: AI can help almost anyone pose as a journalist. That adds up, potentially, to endless competition and blurs the lines on what a journalist actually is now.
Whether they like it or not, almost everyone is using AI these days – every time you ask Google a question.
There is also a potential problem arising from the unscrupulous end of local and regional newspapers.
Businesses strapped for cash and already with hollowed-out newsrooms could be tempted to get rid of most of their remaining journalists and mainly publish the work of AI bots.
How AI has helped the FT
At the other end of the market, publications such as the Financial Times are using AI to investigate stories that would be quite impossible for a human to handle.
The following information on the work of Chris Cook, who leads a group of reporters, engineers, and data scientists at the FT, is first-hand and reliable because I conducted the interview myself for Publishing magazine.
The FT collected 600,000 job ads in Russia – eventually the total got to 1m – using AI systems, and the analysis showed that the Russian economy was indeed overheating, largely because of the invasion of Ukraine and labour shortages because of the needs of the Russian war machine.
The FT was able to confirm, using satellite data from the European Space Agency, looking at 150 defence industry sites, that there really had been a surge in rearmament across the continent, rather than just talk about it.
Cook also told how an individual foreign correspondent, Laura Dubois, in the Brussels office was able to use the FT’s AI systems to track filings by MEPs about their second jobs.
After reviewing thousands of documents using AI, Dubois demonstrated that many of those second jobs involved conflicts of interest with their duties as MEPs.
Telling examples, and naturally, nothing leaves the FT without being reviewed by human eyes. But how widespread is the use of AI around the world, and how is it being used and with what results?
What better to do than ask ChatGPT directly?
The results, covering several thousand words and neatly categorised, were available in less than 20 seconds.
It goes almost without saying that it would be virtually impossible for any individual freelance journalist to assemble such information any other way.
The findings are remarkable. Apparently, around 75% of news organisations globally use AI in some form, and in the UK, over half of journalists use AI weekly.
ChatGPT concludes that AI is now embedded across research, writing, editing, publishing and distribution – which sounds pretty much like the entire chain.
Journalists are routinely using AI to search large documents, summarise reports, court filings, and transcripts, and identify patterns in data.
Although more than 10% of journalists already use AI to generate first drafts. Some reporters use it to suggest ideas and ask AI to come up with stories in their style, but AI rarely publishes alone without human editing.
Opinion pieces are more likely to contain AI material than news stories, although the Associated Press in the US produces thousands of financial and sports reports automatically.
Nearly 50% of journalists use AI for transcription, one-third for translation, and 16% for generating ideas.
AI is used to scan social and data feeds to identify emerging trends or underreported stories.
This is all very interesting and perfectly plausible, though God knows where all these statistics have come from.
To continue, journalists use AI to clean up copy, standardise tone and check statistics. It also helps reporters turn raw datasets into interactive graphics and readable summaries, a practice used heavily in climate reporting, election coverage, and public health journalism.
Is there anything that AI cannot do?
At the cutting edge of technology, AI chatbots are already answering readers’ questions, and voice clones are being used for audio journalism.
There is even AI-assisted investigative hypothesis generation and, in some countries (unspecified), synthetic TV anchors.
Moreover, journalists use AI tools to identify deep-fake videos, fake images and coordinated disinformation.
“Nearly 90% of journalists believe AI increases disinformation risks, so detection tools are critical,” says ChatGPT in a clear case of heads AI wins, tails AI wins.
The system’s conclusion about the future of journalism and AI is the most interesting of all.
AI is not replacing journalists, but it is reshaping their work.
And while you have to be cautious about AI conclusions, AI claims that journalists are becoming “AI-augmented professionals” rather than obsolete ones.
Now that is a conclusion we can all get behind, from humble freelancers to the grandest newsrooms in the land. That’s definitely not a bad haul for less than 20 seconds’ work.
Raymond Snoddy is a media consultant, national newspaper columnist and former presenter of NewsWatch on BBC News. He writes for The Media Leader on Wednesdays — bookmark his column here.