Saturday morning

Good morning! We’re a full house at the moment and for the foreseeable, and the combination of high summer, lots of people staying, lots of cooking, late suppers in the garden with candles sputtering, and a holiday feeling in the air - even if I am physically at home and working - means that part of me is in unofficial holiday mode. You know when you go a bit gormless and vague and just want to lie about barefoot, eating cherries? I have that. I’ve said ‘Oh God, sorry, were you expecting a piece of work from me?’ twice this week.

So I’m thinking this weekend missive might feel slightly different for the rest of summer - I mean, it might not, I never really know until I sit down to type, but I'm thinking it will probably be more a letter that points you to a few things that have caught my eye, or are swirling about my head, than the more usual deluge of satisfying things to consume in one way or another.

I might even take a week off at some point 😱 - it’s literally been years (which makes me sound much more diligent than I am. I never think of this newsletter as work because I have such a nice time writing it, but it does take me all week to think about and shape in my head).

Do you know, I’m absolutely not wishing anything away - this summer is being total heaven thus far - but I am weirdly looking forward to early autumn, new notebooks, sharpened pencils, cooler nights, that feeling of rolling up your sleeves again. A certain briskness. The return of cosiness! I’m itching to show you new autumn clothes, and new autumnal bits for the house, and to chat about brown food.

But it’s too soon to be pointing you to things made of wool. Equally, I don’t feel there’s much point in steering you to a load of summery things, especially things that are now only available in one size and otherwise sold out everywhere. There’s always my shop for any last-minute holiday needs.

So. We will be patient. We will bide our time.

I write these posts for paid subscribers once a week, and for free subscribers once a month. This is a paid post. Have a look at my home page for dozens more like this one. Free posts auto-paywall after 2 weeks.

Substack has linked up with Pangram, the AI program that detects AI (is this not an issue in itself?), and which you can now run on any note or post if you feel like having your suspicions confirmed, or like playing Witchfinder General. Weirdly, you can also run it on your own draft posts. I don’t really understand this - either you’re bonkers and don’t realise you’re using AI (??), or you’re fiddling about trying to make AI read like not AI until Pangram tells you you’re good to go. Neither scenario seems ideal.

Still, I’m pleased about this at one level, although I feel I know AI when I see it (for now) without needing to get all forensic about it. But I also wonder whether it might lead to people being called out publicly for being bad at writing or not having the confidence to write in their own voice (the two aren’t the same). Or for being knackered but still wanting to post, or having children that never sleep, or going on holiday and not wanting to let their subscribers down, or whatever.

To me, these are all forgivable sins. The thing is always to say if you use AI. I think people only feel cheated when work is presented as original and authentic when it just isn’t. As I was saying a fortnight ago, I free-subscribe to a couple of newsletters that are entirely written by AI (according to me) but contain some really good visual ideas, even if they are horribly expressed.

Fun fact: Anthropic used 7 of my books to train its LLM. I don’t use AI for anything I write, or for any aspect of anything I write, like spellchecking or editing. But that’s because I can do those things myself (I started off as a sub-editor, which is being the person that knocks other people’s copy into intelligible shape and, in the olden days, made it fit on the page). If I couldn’t do these things, I expect I probably would get Claude to give me a hand. Why wouldn’t you? Again, I feel a tiny bit sorry for people who are shit at spelling or grammar or ordering their thoughts, but have something to say - they will be denounced by Pangram too, even though they don’t belong to the same category as people who just get AI to write essays that they pass off as all their own work.

I do use AI, though, probably in quite a babyish (or old-ladyish) way. I’ve been trying to make it build an index of all my posts, which would be so useful, but it seems to find that a deeply challenging idea. Still, it’s brilliant at visualising things like paint colours or wallpaper in a room, or what the sofa would look like if it was over there instead, or what would happen if the lighting was different, or what this or that planting scheme would look like, and so on.

It’s also good at decoding ingredients on food labels, or in e.g. expensive skincare - like, why is this £250 when the thing with similar-sounding ingredients is £17? - and at skincare routines if you ever wonder whether product A works optimally with product B, especially where both products contain actives. And you can take a photo of the contents of your fridge and ask AI to suggest recipes, which is especially handy if said contents are tragic.

In case you missed it at the time, there are more helpful domestic uses of AI in this post from last year from Anna Eleri Hart. (Do share any you might have in the Comments).

The main thing, though, and what I feel has got rather lost in the whole debate, is that vast numbers of people don’t mind reading AI ‘slop’. They can’t spot it, they enjoy reading it, and really they don’t care who wrote it, or how. So there’s that.