On not spotting AI writing
Less than a couple of weeks ago I read a post on Substack in support of long form writing. It also expressed concern about the move towards Notes, a feed of short clips, video and endless scrolling.
I agreed with the piece. It said what I had been thinking for a little while. It wasn’t clever or particularly profound, it was stating the obvious and slightly annoying trend of Substack becoming the online places we had left behind. It was a plea to the CEO of Substack to protect what had been built.
Hurrah I though and happily restacked the piece. It already had hundreds of likes and restacks. It seemed to be a popular position.
Today, I find out that the piece was written by AI.
I read an article by a writer I appreciate and respect who determines that the issue is not that people are using AI (although ugh) but that readers cannot tell the difference.
I was one of those readers.
I’m confused now, honestly. I couldn’t tell. I didn’t occur to me. I certainly didn’t run through a (potentially inaccurate) AI detector before restacking. I have now and it reports 100% AI generated.
So now I’m swirling in a pool of why not. Why couldn’t I tell? Why did I find the writing compelling enough to restack? What skills am I lacking? Am I actually just stupid? Should I delete the restack? Should I publish some kind of apology for being so daft? Did I mislead anyone else to restack this post? Am I supposed to run every article I like through a checker tool now?
Because it matters to me. I don’t want to promote AI writing and accounts that use it. ( This is mostly about the common concerns about energy use, the unrestrained power of AI building companies, the long term consequences of AI use on human brains and a love of genuine human creativity.)
Perhaps most painfully, how can I even consider becoming a writer if I can’t tell when something is written buy a machine?
Linda Carroll writes about the use of AI writing. She frequently talks about how humans can’t tell the difference. I want to be like Linda, knowing straight away. But do I have to go to every post with a cycnical side eye, waiting to spring, “Ah ha, you’re not a real writer.” at the ready.
That’s horrible. But apparently the truth is that AI writing is everywhere and I’m not smart enoughh to recognise it and that makes me a part of the problem. So what now? Is anyone running Substack courses on how to recognise AI writing? Are we still talking about em dashes?
Kore