It’s clear the next big leap forward in journalism upskilling is in using generative AI and journalism educators need to catch up fast.
Media organisations, including Reach, Newsquest and the BBC have adopted AI technologies to refine and personalise content, create audio versions of written content and identify trending topics. Reach says it uses AI in “each step of its storytelling process — sourcing, creation, editing, publishing, distribution, amplification, and assessing story effectiveness“.
Reach has its Guten AI Rewriting Tool to ‘rip’ content (rewrite a story for different audiences) allowing the company to “do things that we wouldn’t otherwise be able to do”, according to editorial director Paul Rowland.
It’s an optimistic view of the potential of AI, freeing up reporters to hit the streets and seek out stories while software takes care of the boring stuff back at the office. In a similar mould are News UK’s News Transcribe, a speech-to-text tool designed to transcribe interviews, and PA’s use of AI to analyse Freedom of Information (FOI) requests.
While the industry wrestles with the legal, ethical and even existential considerations of using AI in journalism, it’s still roaring ahead with deploying technology. Like the advent of the internet, it means journalists new to the industry and those still training need to add a whole raft of new skills to their armoury.
As a journalism educator, it’s my job to figure out how we teach students about using AI while conveying the risks and red lines of deploying the software in an academic or professional setting. But one of the ways in which lecturers in more vocational subjects build modules and programmes is to look at what is happening in industry and work backwards from there.
Speaking to practitioners can be invaluable, but job adverts are a quick and easy way to follow trends and see what hard skills are in demand – they show that young journalists will be expected to know how to pump a Large Language Model (LLM) such as ChatGPT for research, rewrites and refinement. In this way using AI isn’t that far from using a search engine, but caveats apply here. ‘Garbage in, garbage out’ is an old coders’ saying; the same is true for search and AI queries.
An overlooked parallel with how to use AI in journalism is understanding how to use search modifiers in Google. With knowledge of some elementary coding — usually little flourishes of punctuation in a search query to force a search engine to include, exclude or refine results; to search specific websites or file types — anyone can use these sites to massively enhance their journalistic research.
In the same way it’s important to understand how LLMs interpret your meaning. You don’t have to look far for an example of AI committing heinous errors in journalism and you won’t have to ask many questions of a chatbot (especially if scraping facts from the web) before identifying answers that are clearly incorrect or getting responses that aren’t what you wanted.
But where search engines are largely dumb (despite accelerating their own AI features into search), you can have a de facto conversation with ChatGPT. This dialogue, or even dialectic, is where LLMs can be invaluable, or worse than useless, and it’s where journalism educators can give their students useful direction.
The quality and clarity of these prompts is vital in getting the best out of most AI software. As with a poorly written press request or FOI, the nature of the prompt is key in getting something useful out of an LLM. ‘More formal’, ‘write for an audience reading The New Yorker, ‘show citations and use data from 2025’ are all commands an LLM will respond to.
Direct it to write to a word count, in a specific style or structure or to cater for a geographic location, and most are up to the job, producing something readable and accurate, refining and honing your work until it fits your requirements. Ensuring journalists can spot inaccuracies or problematic answers is another area where guidance from educators will be vital in the inevitable roll-out of AI.
But understanding what this software can do, its possibilities and limitations — and the ethical boundaries — is key, not simply for the journalist of the future, but the journalist of today.
For journalism lecturers and trainers everywhere there’s a new show in town (yes, another one, sorry about that) and you need to know about it. If you don’t, your students will do it themselves — only without the guard rails.
Email pged@pressgazette.co.uk to point out mistakes, provide story tips or send in a letter for publication on our "Letters Page" blog
