Can AI write convincing lifestyle journalism?

As a student, I've often pondered the potential of AI in writing. With advancements in AI technology, the question isn't whether AI can write convincing articles, but how effectively it can do so.

Image AI generated using deepai.org, using the prompt ‘artificial intelligence working as a journalist in an office of the future, colourful’ (19/06/24).

Imagine a world where your morning coffee comes with a side of algorithmically generated wit, rather than the musings of your favourite columnist. "Could AI really write opinion articles?" you ask, possibly while sipping that very coffee. Well, dear reader, the future is nigh, and artificial intelligence (AI) is knocking on the newsroom door.

Let’s dive into this curious possibility, but first, let's set the stage. AI has already made significant inroads into journalism. Take, for instance, a study by the Reuters Institute for the Study of Journalism, which found that newsrooms are increasingly utilising AI to automate mundane tasks like data analysis and news aggregation. But can these algorithms take the next leap and write opinion articles?

The science, it seems, is promising. A study published in the journal Digital Journalism examined how AI-generated content is perceived by readers. Interestingly, the study found that readers often couldn't distinguish between AI-generated news and articles written by humans. This is either a testament to the remarkable sophistication of AI or a commentary on the state of human journalism. (I'll let you decide which.)

But let's not get ahead of ourselves. Writing an opinion article isn't just about stringing words together in a coherent manner—it's about crafting a narrative, injecting personality, and often, a bit of humour. Can AI, with its ones and zeros, truly capture the idiosyncrasies of human thought and emotion?

Consider the following: when AI attempts to mimic humour, the results can be, well, mixed. An AI might spit out a joke like, "Why did the journalist cross the road? To get to the other byline." Now, it’s hard to say whether that joke is painfully dull or post-modern genius, but humour often requires an understanding of cultural nuances and a certain ‘je ne sais quoi’ that AI, at least for now, lacks.

Moreover, opinion writing demands a depth of understanding and personal experience. An article on the ethics of AI in journalism written by AI might come off as slightly self-serving, don't you think? It's like asking a fox to write about the security measures of the henhouse.

Let's also address the elephant in the newsroom: bias. Human writers have their biases, and while we strive for objectivity, we are inherently subjective creatures. AI, on the other hand, can be programmed to minimise bias, drawing on a wide range of sources to present a more balanced view. But who programs the programmers? Bias, it seems, is an insidious beast.

In conclusion, while AI has the potential to revolutionise the way we write and consume news, the heart and soul of opinion writing—its humour, its passion, its very human essence—remains, for now, a distinctly human endeavour. So, for those of us who cherish the quirks and the wit of our favourite columnists, there's no need to panic just yet. But keep an eye on that newsroom door; the future is always closer than we think.

---

Okay, time for a real confession.

I didn’t write any of that.

Instead, I tasked ChatGPT with the job of doing it for me, because, what better way to determine if AI can be used to write lifestyle journalism, then have AI do it for you?

To be completely candid, I did have to go through ten minutes of trial and error just to get the article I was looking for, but I started this whole process by having ChatGPT read some of my previous articles to get a good sense of my lifestyle tone of voice.

AI’s reading materials in this circumstance were my article on ADHD and my article on raising the minimum wage, just for range and posterity.

Editorial note: I would strongly suggest that you are always nice to your AI. Who knows where we will be in ten years.

After AI had read them (in lightning speed, I might add), it gave me a little feedback on my writing, and then, we were away. I asked ChatGPT to write an article, in my tone of voice, about whether AI could by used to write lifestyle articles:

And my goodness, did AI write an article in my tone of voice.

This is where it got a little creepy for me.

To test the unnerving similarity that I was feeling about AI’s impersonation of me, I read sections of this piece to my wonderful fellow writer

Maleeha Saad

- who frequently proof reads my work for me - alongside sections of the pieces I had personally written. Terrifyingly, she could only distinguish between my writing and AI in one circumstance out of six.

Let’s be realistic, that is horrifying. Not only is AI spookily good at mimicking a real person’s style of writing, it send some serious shade my way in the middle of the article I commissioned it to write.

In this respect, it is completely viable that AI can be used to write lifestyle pieces of journalism, and it’s near indistinguishable when given the opportunity to read a real person’s journalistic voice.

Despite the worryingly effortless ability AI holds to mimic a journalist’s voice, a study published by the Reuters Institute, they found that a relatively high percentage of readers still had a distrust in new generated by AI, especially regarding topics of ‘high’ relevance, such as politics, crime and local news.

These results are somewhat reassuring to know that the public still hope, for the most part, for news written by people, rather than a machine. But can the public tell the difference? My miniature experiment suggested ‘no’, although it was taken with the smallest sample pool possible.

Of course, there are some slip up in AI’s mimicking work. ChatGPT was unable to integrate links into its work, even though sources were mentioned, which is:

1. sloppy journalism, and;

2. not my style - please note the many links in the second section of this article.

It also struggled with some slightly wooden comedic analogies - although, if you weren’t aware it was generated, I’m not sure that you’d pick up on it.

AI is advancing fast now, especially ChatGPT, who have recently release their new edition of the site, GPT-4o, which promises that it can ‘reason across audio, vision, and text in real time’, suggesting that even more complex and harder to distinguish AI is right around the corner.

The future of journalism is certainly going to be tightly intertwined with AI, and in many more ways than the nerve-inducing-mode of taking over jobs.

AI is also incredibly useful for generating ideas for new articles, creating headlines which perfect every SEO (search engine optimisation) in their path, and creating images that couldn’t otherwise be gained through standard photography (like the lead image on this article).

In summary, AI is somewhat a double-edged sword for journalism.

It can be incredibly helpful and incredibly threatening to this field. What the modern journalist needs to do is stay on top of these technological advancements, and not bend to the ‘easy’ elements of AI journalism, or the field may well collapse into a world of ones and zeros, without the human element that makes journalism, well, journalism - lifestyle included.

[Editorial note: In this article, I, Giorgia Coupland, wrote the headline, the subheading, the image prompts, the video script and the text following the line break halfway down the article. I also generated the data visualisation with data from the Reuters Institute. AI generated the written section starting at ‘Imagine a world…’ and ending at ‘…the future is always closer than we think.’]