Seán Moncrieff: Can ChatGPT write this column for me?

Do I have to explain what ChatGPT is? If you’re unfamiliar, you can google it. Because, in a way, it’s the new Google.

When you log in to the new artificial intelligence chatbot, it’s quite ordinary. I would have preferred some momentous-sounding intro music, just to prepare me to face this sci-fi God. Hello Dave, everything’s running smoothly. But it’s no HAL.

Of course, I asked about me, and it told me that I was born in Birmingham (I wasn’t), that I grew up in Wexford (nope) and that I’m the author of The Sneaky Chef: How to Cheat on Your Man (In the Kitchen), which old-fashioned Google says was written by an American woman called Missy Chase Lapine. I would never cheat on my man, in the kitchen or anywhere else.

I was mildly insulted. Not so much because it got a lot wrong: it was more a sense that this supposed totem of world-changing human ingenuity couldn’t be arsed to look up Wikipedia.

Seán Moncrieff: Psoriasis is one of the flaws that make me who I am

---

Seán Moncrieff: To the British, the Irish aren’t really foreign

---

Sean Moncrieff: Why are Irish novelists so successful? Perhaps because they’re telling us the story of ourselves

---

Seán Moncrieff: We would prefer to destroy the world rather than stop being ourselves

---

This isn’t my first encounter with artificial intelligence; nor yours either. Those dialogue boxes that often pop up on websites, usually featuring a smiley human head, are also powered by AI. You’re trying to find out if that parcel has been lost in the post or why you’re suddenly €5,000 overdrawn, and the AI will politely tell you what you know already, while maintaining the fiction that it is somehow helping you.

I’m far from an expert in these matters, but, rather like ChatGPT, that won’t stop me expressing an ill-informed opinion

But this is the future. AI is already being used in various manufacturing processes, the sorts of things that require doing a lot of sums quickly. Which is fine. But it’s also encroaching into the muddier area of human interaction. It has a role in deciding who can get bail in parts of the US court system. It’s used in facial-recognition technology. For sifting through CVs. I’m far from an expert in these matters, but, rather like ChatGPT, that won’t stop me expressing an ill-informed opinion: in its human-facing interactions, AI gets things wrong. A lot. It absorbs the biases of its programmers, and it seems to lack an ability for nuance, context and empathy – the fundamental things that make us human and influence how we act.

Yet so many companies and government agencies are keen to adopt it, perhaps in the full knowledge that, in many cases, it won’t be good as a real person. But it will be quicker. And cheaper. The near future will be a bit more inefficient, but a lot more profitable.

Not everyone is a fan. A lot of scientists and tech bros are warning that we should pause AI research in case we inadvertently trigger a Skynet situation. And even at the dumber end of AI, there is concern that students will use ChatGPT to do their homework for them or that it could be put to various nefarious uses. One user talked it into giving the recipe for Napalm.

[ ChatGPT poses an existential threat, and the window for gaining control over it is small ]

So, I gave it the only test that I’m qualified to give: write an Irish Times column in the style of Seán Moncrieff.

The drivel it vomited up. Banal and cliche-ridden. No mention of Herself or the number-children. No woke virtue signalling. It kept using the term “musings”. On the first attempt, it opened with 70 words about the weather. So, I gave it another chance, and told it to mention the kids. It claimed that one of them taught me to play Minecraft. On the third attempt, I asked what I might say about artificial intelligence.

It was a stream of platitudes, but ended with these words: let’s all stay curious and open-minded as we navigate the brave new world of artificial intelligence. The reference to a novel about a dystopian future might have been accidental, or ignorant, or vaguely threatening. It is possible that artificial intelligence is just pretending to be stupid. In which case, it’s already too late.