I’m usually not someone who gravitates towards pessimism or endless talk about the decline of civilisation. And I’m not about to do that here either. This isn’t going to be another post about AI suddenly becoming sentient and wiping out humanity.
But…
Over the past few months, I’ve started noticing that AI is already damaging some very human things. A few examples.
I have a very mixed relationship with LinkedIn. On the one hand, it sends a lot of traffic to this blog, which is great. On the other hand, more and more of the content I see is clearly written with or by AI. I’ve had a Pangram account for a while already, but these days I often don’t even need it anymore to recognise AI. One particularly popular format currently: one starts with a tiny anecdote (“I bought a dress today”) followed by a sweeping reflection on some major global issue.
Speaking of Pangram, the thing has made me surprisingly insecure about my own writing. I wrote a lot during July and regularly ran my own texts through AI detectors. Mind you, I knew perfectly well that I had written them myself. Some of those texts even dated from before 2022, so AI literally couldn’t have written them. Apparently, though, some sentence structures I happened to use back then have since become popular among AI models. I’ve even caught myself avoiding em dashes for quite some time now – even though I used them all the time in the past.
An even stranger example: I had a paragraph that three different people edited independently, after which a professional editor revised it once more. That final version was suddenly labelled as AI-generated, whereas my original version wasn’t. To be clear, I do use AI. But not in the sense that AI writes my texts. Grammarly suggests changes while I’m writing. I ask Claude or ChatGPT for feedback. I use ChatGPT or Google Translate to help with English versions. But how often have I received feedback telling me to remove my quirky humour or rewrite sentence structures that I actually liked? Which brings me to a third trend.
Our language will become poorer. I’m not just talking about what happens if people stop writing for themselves. The technology itself seems to push us towards the same bland style. The irony is that the AI companies know this too. I wrote about it earlier on this blog, based on a Nature paper. The supply of high-quality training material is drying up. As a result, the major AI companies are now almost literally buying up parts of our cultural past, scanning them and destroying the originals.
And while writing that previous paragraph, another frustration came to mind. AI has ruined search engines and search functions. No, really. Google’s search results have become depressing. Searching within WordPress has become unnecessarily time-consuming. Even finding an email in my own mailbox can be an ordeal. AI was supposed to make all of this better, but I can’t even begin to estimate how much time I’ve lost because of it.
Okay, now I’m definitely ranting, so it’s probably time to stop. I won’t even get started on the environmental impact, the concentration of power, or the possible economic consequences. This blog is already full of posts about AI and education. Oh, and I won’t start talking about AI and music either, even though there has just been an important court ruling on that front.