Bored of it

…

Every article glorifying it.

Every article vilifying it.

Every pub conversation winding up talking about it.

People incessantly telling you how they use it.

I feel dirty using it.

You know what I’m talking about, even though I’ve not mentioned it.

…

If you don’t know what “it” is without the rest of the context, maybe read the rest of Paul’s poem. I’ll wait.

As you might know, I remain undecided on the value of GenAI. It produces decidedly middle-of-the-road output, which while potentially better than the average human isn’t better than the average specialist in any particular area. It’s at risk of becoming a snake-eating-its-own-tail as slop becomes its own food. It “hallucinates”, of course. And I’m concerned about how well it acts as a teacher to potential new specialists in their field.

There are things it does well-enough, and much faster than a human, that it’s certainly not useless: indeed, I’ve used it for a variety of things from the practical to the silly to the sneaky, and many more activities besides ¹. I routinely let an LLM suggest autocompletion, and I’ve experimented with having it “code for me” (with the caveat that I’m going to end up re-reading it all anyway!).

But I’m still not sure whether that, on the balance of things, GenAI represents a net benefit. Time will tell, I suppose.

And like Paul, I’m sick of “the pervasive, all encompassing nature of it”. I never needed AI integration in NOTEPAD.EXE before, and I still don’t need it now! Not everything needs to be about AI, just because it’s the latest hip thing. Remember when everybody was talking about how everything belonged on the blockchain (it doesn’t): same energy. Except LLMs are more-accessible to more-people, thanks to things like ChatGPT, so the signal-to-noise ratio in the hype machine is much, much worse. Nowadays, you actually have to put significant effort in if you want to find the genuinely useful things that AI does, amongst all of the marketing crap that surrounds it.