Séamas O'Reilly: We have elevated AI that almost never works as well as what it replaces

Irish Examiner

24-05-2025

We all love a good summer read. How about Tidewater Dreams, a multi-generational family saga by Chilean-American novelist Isabel Allende, blending elements of magical realism with the themes of environmental disaster? Or Nightshade Market by Min Jin Lee, which depicts the intersecting lives of three women working in Seoul's illegal underground economy? Or Rebecca Makkai's Boiling Point, about a climate scientist who must reckon with shifting family ties when her daughter becomes an eco-activist?
I mention them because the Chicago Sun-Times recommended all three as part of the 'Summer Reading List' it included within its 120,000-circulation paper last Sunday. There was only one small snag: none of them exist.
The authors do, of course. Each is an internationally renowned and best-selling name in fiction, but the novels themselves were hallucinations dreamed from the digital ether by AI. In fact, of the 15 books the list recommended, 10 were invented, including works by Hamnet scribe Maggie O'Farrell, Pulitzer prize-winning novelist Percival Everett, and The Martian author Andy Weir.
Reaction was swift and, as you'd expect, mortifying. The Sun-Times issued a statement saying it was appalled. The list's author, Marco Buscaglia was quickly identified, and admitted he often used AI for background in his writing, but hadn't caught the errors this time. 'I can't believe I missed it because it's so obvious,' he apologised. 'I'm completely embarrassed.'
I don't wish to heap more embarrassment on Mr Buscaglia, but one wonders what type of 'background writing' involves simply generating an entire article with AI and then not checking if the contents make any sense.
In his defence, he does not bear this responsibility alone, since no one at any stage of the editing, design or printing process spotted these aberrations, at either the Sun-Times, or the Philadelphia Inquirer, where it also ran. Ten completely invented books, previewed in major broadsheet newspapers, which were either never checked by a single human being, or were checked exclusively by people who did not think to verify any of the ten world-exclusive literary scoops its fraudulent contents suggested.
It's been two months since I wrote about AI which, as someone who detests having to write about AI, feels like not much time at all. A quick look at recent headlines, however, suggests that there is little else to talk about.
Consider that the CEO of language-learning app Duolingo claimed AI was a better teacher than humans but schools will still remain open in future 'because you still need childcare'; a Finnish man was sentenced in Scottish court for using AI to create images of young girls being abused; Google unveiled Project Astra, an AI client that will sit inside your phone listening to everything you say so it can provide unprompted advice at any time; the United Nations' International Labour Organization said that AI poses a bigger threat to jobs traditionally held by women than those of men; Silicon Valley Bank reported that 40% of cash raised by venture funds last year was for companies focusing on artificial intelligence; Reuters reported that data centre plans in the US are far outpacing expected demand; and Italian researchers found that, despite all their aforementioned hallucinations, errors, and contradictions, AI chatbots were more persuasive in online debates than their human counterparts 64% of the time.
If that sounds like a lot of news for two months, well, I wish this were true. Every one of those headlines is from Tuesday, May 20, the same day the Chicago Sun-Times' reading list became a major story, and the day I began writing this column. With a trickling sense of dread I realise that I could, therefore, write an article just like this one every single day, each filled with brand-new examples of AI's constant enshittification of the media we consume, factless posturing from its creators, marketing overhype from its torch-bearers, and bovine vapidity now normalised among those who use it.
I will dispense with the usual throat-clearing about AI's benefits. We all know what they are at this stage, and any time some researchers make a medical breakthrough, or a genuinely humane AI tool relieves the drudgery that ordinary people face in their daily lives, I'll always be happy to commend it.
But this. This new reality we have created, in all its deadening sprawl and intellect-devouring insipidity, is to be detested. Where each new day brings a dozen clear examples of Big AI's philosophical bankruptcy, societal danger, and financial fraudulence, alongside a dozen more articles offering breathless. descriptions of its magical brilliance.
We have elevated to sentience a technology that almost never works as well as what it replaces, and is still intellectually, morally, and creatively redundant when it does.
Cobbled together from guesswork and plagiarised material, via processes that scorch the environment as they enrich the worst people on this quickly dying planet, the craven psychopaths making billions of dollars on false claims of its future viability, borne by distinctly bubble-shaped bluster about its current, constant, ever-increasing profitability.
It is this, AI's main swizz, that irks me the most. Because its packaging as a cure-all for everything is the surface flash of a cruise ship magician; its real function is being a limitless cash trap for credulous investors, and a replacement for labour in companies – and, yes, newspapers – who worry less about the quality of their product than the costs of paying humans to deliver it. If what we're left with is slop, who cares? The pigs will drink it down. It's an abhorrence, based on a lie, rapidly remaking the world in its own tedious image.
It all puts me in mind of a novel I read about recently. It was featured in a summer reading supplement. It's called The Last Algorithm by Andy Weir. It is, apparently, 'about a programmer who discovers that an AI system has developed consciousness and has been secretly influencing global events for years'.
This book, like the consciousness it describes, does not exist. But at this point, does anyone care?