The Gist: The AIrish Times

This week saw a media novelty, as an AI generated article ran in the Irish Times. This is The Gist.

The Gist: The AIrish Times

This week saw something new in the Irish Times. As the paper’s editor later acknowledged, the newspaper unwittingly published an opinion article at least partially written by an AI chatbot, along with an AI generated photo of a non-existent blue-haired Ecuadorean author.

The involvement of this technological novelty was the focus of much of the subsequent debate, with the newspaper identifying it as an issue to be addressed in future commissioning. (Presumably, henceforth all opinion columnists will be treated as potential fae changelings, with Fintan O’Toole asked to show his teeth and hands and touch a piece of cold iron before his column is accepted).

Not addressed by the Editor’s note, because it was a far more difficult issue, were the habits of thought which saw an article of the sort capable of just being generated by a chatbot being published in the first place.

This seems to have been the hoaxer’s low ambition- to prove that the Irish Times was so eager to give voice to under-represented opinions that it would publish absolute by-the-numbers culture war nonsense.

White Musk

Certainly it was the gleeful squeal from the White Musk right, and, in a self-exculpatory way was also the paper’s after the fact position. The Editor’s note defiantly ended with a pledge to keep publishing under-represented voices, (such as the previously silenced non-existent person population).

What it actually revealed was something deeper, and in the long term more concerning for all of Irish media.

The article was published for the clicks. It got the clicks, because its contents were provokingly stupid. When it was published, it was the second most popular piece on the site.

The most revealing part of the Editor’s note was the blithe assertion that this is what opinion pieces are for- “The aim each day is to come up with a blend of thought-provoking columns that inform, stimulate and lend a fresh perspective on a current issue.”

Why was wearing fake tan a “current issue”? Well, it wasn’t. Literally nobody in the country was thinking the things generated by the AI chatbot. We live in a country where seeing the sun becomes a topic of surprised conversation amongst strangers. But that sort of thing had been a successful culture war topic elsewhere, and this was a chance to tap a fresh new vein of rageclicks. “It made an argument that has been aired in other countries but related it to the Irish context.” as the Editor’s note put it.

ClicksGPT

It was the content, not the hoax, which was the real breach in trust with readers. Whether the author made their argument in good or bad faith, it could only be published for indefensible reasons- not to give voice to the voiceless but to angry up the blood of the readers. The hoax/AI elements simply stripped the paper of its usual defence for publishing this sort of thing. They couldn’t argue it was a legitimate point of view they just happened to be harvesting clicks on, because suddenly the blue haired immigrant they were shielded by winked out of existence. Now there was nothing standing between the text and its promoters.

Newspapers everywhere publish opinion columns to give a platform to chosen arguments (and the people who make them). There is frequently an internal understanding that these columns are quite unlike the content produced as news, where being accurate and factual are, in the Irish Times as elsewhere, critically valued. Facts, as the old Guardian Editor’s comment goes, are sacred.

But though the world is full of facts, people aren’t filing cabinets. We make sense of the information presented to us by placing it within broader narratives about what’s going on in our world.

This is the real purpose of an opinion section and the real risk posed by one which wrongly thinks it is there to “provoke” as a primary purpose. It provides narratives and stories that help readers make sense of all those facts so carefully assembled in the front of the paper.

There is huge power in narrative building. A false narrative can become so accepted through repetition that it can lead to profound errors of public policy and worse. Understanding is built on facts, but it doesn’t grow by them alone. Stories are stronger than mere facts, which makes it so important not to reproduce harmful and divisive ones just because you saw it get clicks elsewhere.

From Brexit to Covid denial, from the invasion of Ukraine to the climate crisis vehemently denied by voices given space in newspapers and other media for decades, the platforming of bad ideas and opinions and the reproduction of harmful and wrong narratives has threatened lives, jobs and polities.

Wavin Pipers

I trust the Irish Times to always attempt to get its facts right (nobody’s perfect, but I believe that’s the aim of every news report). But ever since its previous opinion editor defended publishing a far-right glossary of abusive terms, written by a proponent of right-wing provocation, it has been clear that the paper has gone wrong when it comes to understanding its own role in helping readers make sense of the world.

Whether written directly by an existing human or generated by an AI bot at the prompting of a hoaxer, the threat from platforming bad faith arguments is the same. As in economics a debased currency will overwhelm any remaining pure coins in circulation, so in the realm of platforming opinions, bad takes drive out the good. Not recognising the difference between the two is what loses a media outlet its readers’ trust.

We haven’t needed AI to produce debased opinions in the past- we’ve always been able to find humans ready, willing and able to generate them on demand in exchange for money and/or influence. The only difference the intervention of an AI chatbot makes is to lower the barriers of entry so as to ensure those open sewers may gush unceasing.

We need to be able to tell our stories to each other. But the role of an editor is not to act as a journalistic Wavin pipe, dumping undifferentiated contents on the reader. Not all stories deserve to be heard, either because they take no account of the facts of the world or because they seek to harm others through bad faith argument.

Finding and promoting the stories that help improve our lives, rooted in fact and tinged with human insight is the promise of an opinion section. Trust builds the more often that promise is kept.

💡
Some interesting things I've read on AI recently: Wired on finding out ChatGPT has been stealing your work. Interesting new angle on the cultural impact of a post AI world in a podcast discussion. And, just to confirm, ChatGPT will generate an essay in the style if the Gist if you ask it to, so at least we know we have one famous reader. Even if it is a non-sentient chatbot.