Unmasking the AI frauds

When public commentators - such as a former colleague of mine - try to pass off slop content as their own work, do they deserve to be exposed?

You probably won’t agree with all of Mr Embery’s policy prescriptions, but he will force you to think outside your usual political grooves — Wall Street Journal

Photo: deepak pal, via Wikimedia Commons

I KNOW VERY little about artificial intelligence (AI). I can scarcely tell my Copilot from my ChatGPT, or whatever they’re called, because I never use these tools. Or I haven’t done so yet, at least.

People might find that a little difficult to believe in this day and age – especially coming from someone who writes for an (albeit small) public audience – but the truth is that I have never been quick to embrace new technology (I’d be too embarrassed to tell you which model of iPhone I’m currently using).

I did once download an AI-generated image to illustrate an opinion piece I had written, but that’s about the extent of my engagement with this new capability.

When I research stuff, I tend to stick to more tried and trusted methods, such as websites, books and magazines. I dip into Wikipedia occasionally (I have spent countless pleasurable hours just hitting the ‘Random’ button again and again) but never take it as the final word on anything.

I have a constant nagging fear that something I publish will contain an error so egregious that my credibility as a writer (such as it is) will forever be destroyed.

When I wrote my book Despised (still available from all good bookshops, by the way), I spent countless torturous hours, day after day, fact-checking my own text. (I’m pleased to say that, though plenty of opponents attacked the arguments I made in the book’s pages, nobody has ever identified a factual inaccuracy.)

All of which provides a bit of context for what follows.

I used to work with a guy. Like me, he was a senior trade union official. Let’s call him Martin.

Martin has run his own account on ‘X’ (formerly Twitter), a platform on which I am also active, for many years. We never interact these days.

When we worked together, Martin wasn’t exactly the greatest wordsmith. He had strong political opinions – most trade union officials do – but was, frankly, terrible at expressing anything in written prose. His emails and letters were generally a mess, and we rarely permitted him to draft communications intended for circulation among the membership (or, if we did, we would invariably need to spend a good deal of time redrafting them).

I don’t say any of this to be unkind; it’s just the way it was.

At any rate, I recently spotted one of Martin’s posts on ‘X’. It was a reply to a politician. The post was comprehensive, grammatically sound, and the height of eloquence and fluency.

‘That can’t be him,’ I thought. But it clearly was.

So I had a nose through his other recent posts. They were pretty much all the same: expressive, articulate, and free from grammatical errors – the very opposite of the stuff he used to churn out for the union.

Now, even someone such as I, with a limited understanding of AI, could see what was going on: Martin was obviously using an AI tool to create the posts.

But what I also noticed was that, since he had been using AI – his old, clunky, pre-AI posts remain visible, so the precise point at which the change occurred is obvious – Martin’s interactions had spiralled. His posts used to attract a smattering of reposts and ‘likes’; now they were pulling in hundreds – sometimes thousands.

The whole experience got me thinking. What are the ethics here? Is it not wrong for someone to pass off commentary made in the public arena as their own, when in fact it is no such thing? Isn’t there even more of an obligation to be up front about this when the commentary begins to attract significant engagement? Aren’t readers entitled to know they are being served AI slop rather than a genuine piece of written prose? Would it be right to expose those who mislead in this way? Should I use my own reach on ‘X’ to draw attention to Martin’s deceit?

Or should we look at AI as a helpful tool for those who might otherwise struggle to articulate their thoughts and opinions in discussion – a sort of helping hand to provide them with a level playing field in the theatre of public debate?

Not having been born with the gift of wordsmithery, is Martin a bad person for enlisting some assistance in getting his opinions – which I know he holds dearly – across to a public audience? Perhaps not. (Though he was certainly dishonest when, after being challenged by an interlocutor who accused him of the very thing he had clearly done, namely asking AI to write his posts, he firmly denied it. He has also now started publishing full-length articles on ‘X’ under his own name – though again obviously drafted by AI. In my opinion, that, too, crosses a line.)

But perhaps, in general terms, we should look upon the utilisation of AI writing tools by ineloquent individuals as a sort of ‘reasonable adjustment’ – an aid designed to ensure they aren’t placed at an undue disadvantage as they go about their lives. And don’t celebrities employ ghostwriters to pen their autobiographies, almost always without credit? Isn’t that practice deemed permissible by almost everyone?

Joe Weisenthal, a Bloomberg journalist, recently wrote:

Unfortunately, I think that, in the near future, not using LLMs [advanced AI models] to write for you will be like someone refusing to use Google Maps for directions in a new city. A bizarre idiosyncratic choice that’s just completely incomprehensible to the vast majority of people.

Maybe he’s right. But if writers do succumb to such temptation, I feel they have a moral obligation to at least be open with their audience about it. Else they could hardly complain if, in the event their methods are uncovered, their readers end up feeling hoodwinked.

I don’t want to come across as sanctimonious, but I can’t ever imagine a time when I would be persuaded to allow an AI tool to write something on my behalf. The whole exercise would seem profoundly alien to me – no different from asking an artist friend to paint a beautiful watercolour and then mark my signature in the corner (I can’t paint for toffee, incidentally).

None of this is to dismiss the phenomenon of AI or deny its capacity to do good. Though I may be slow to embrace new technology, I am certainly no Luddite. AI is plainly here to stay and will, just as the internet did before it, doubtless enhance our lives in all sorts of ways. (I suspect, mind, that the forewarnings of its likely negative effects – laying waste to large swathes of white-collar industries, for example – may also, in the long run, prove to be accurate.)

But anyone who releases words into the public domain – especially those, such as Martin, who attract large numbers of readers – surely has a duty to be upfront about the origin of those words, including where they have chosen to rely on AI.

Whatever you think of the quality of my prose, dear reader, you may be assured that it is crafted by my own hand. I don’t know if that is anything worth boasting about. But the day it changes is the day I give up writing.

---

A reminder that you can follow me on ‘X’: @PaulEmbery