Interrogative questions for the future of news

The Five Ws are the pillars on which a news story stands. Can generative AI answer their stringent demands?

by Rob Corbidge

Published: 14:47, 29 June 2023

Rob Corbidge is Head of Content Intelligence at Glide Publishing Platform, applying the latest knowledge about advances and ideas in the publishing industry to our own product and helping clients get the most from their content.

Typically a good news journalist will have the Five Ws hardwired into their very being: Who, What, When, Where and Why are the backbone of any reliable reporting. Some of us, myself included, do like to add "how" - everyone wants to know if the candlestick or the dagger was the murder weapon, right?

For this excellent set of interrogative tools, a set made up entirely of simple words, we have Aristotle's Nicomachean Ethics to thank. It's important to remember that as things change, so they stay the same, particularly in this age of apparent technological abundance.

The beauty of this toolkit is that not one of the simple words can be answered with "yes" or "no". Once utilised, this set of questions will unlock everything a news story requires to be built into a form that the reader can understand and digest. The absence of an answer to any of these can serve as a part of the story too, yet it should always be addressed.

Very much as a police investigator does, a news journalist establishes facts.

So can generative AI supplant news reporters? The simple answer is "no". Or certainly not for news in the very sharpest of senses - reporting of an actual dramatic thing that has just happened. 

Generative AI requires seeding to produce content, and that seeding can only be provided by reports produced by humans, whether that be via direct eyewitnesses, a reporter at the scene, and/or collective reporting perhaps via social media etc. 

Success in AI news reporting would then be simply a case of "fastest plagiariser first", building upon the aforementioned human-generated sources and seeds. Isn't that much like online news reporting now? There's no such thing as an exclusive that's not behind a paywall, and even then it's tricky to stop people just stealing your stuff - especially when you might even want them to steal a bit of it, as long as they do the courtesy of an attribution. 

In news production, AI generative content has a role to play. It could potentially pull in what we call "background" - a collation of similar recent incidents following a fatal train crash, by way of example, at a faster speed than a single human reporter, or even a reporter/editor team. 

A formal statement situation could also lend itself to generative AI, for instance being able to peruse the pre-circulated statement given before a Finance Minister makes a speech would enable the AI to (we hope) accurately generate a summary and perhaps even comment on its accuracy. I'd still rather see have a reporter tell me if the Finance Minister has sweat on their top lip though.

Whatever the exact use case, it would be naive to ignore some publishers wishing to reduce headcount and save money on the back of such capability.

It's yet another battle of human curation versus AI scale.

Possible innovations for AI in news could be along the lines of generative AI analysing raw news footage and then producing a voiced or written narrative of what is in the video. Obviously once again, it depends what content the system has been trained on. One man's terrorist is another man's freedom fighter and it's easy to imagine that a Russian government trained system would produce a startlingly different narrative to a Ukrainian government trained system when viewing the same piece of combat footage. Could you say that of reporters? Yes, absolutely. Most reporters are answerable to reality eventually, at least.

In among all the potential disruption (did someone say automated churnalism?) caused by generative AI brings the possibility that news reporters may need to add a larger degree of writing skill to their professional assets, and in doing so make their own content more distinctive and more agreeable to their intended audience by way of style. 

There's an inherent issue here in that anyone who has worked in a newsroom knows that someone who is an excellent reporter and comes back with the dynamite quotes every time might not in fact be be a great writer, and conversely your in-house crafter of the mot juste might not be your preference to have at a murder scene.

Uniformity of style adds also adds authority to news. Imagine a Reuters news report featuring some wild question marks and a few emojis peppered in, because that's the reporter's style. Distinction through unique format, such as Axios for example, can work. Yet formats have their limitations.

At the extreme end of predictions for generative AI news comes the anticipation of a tide of plausible gibberish - things that sound like they could be accurate but are factually flawed. 

The tide, if it comes, won't wash away the five Ws, but it might obscure them for a time.