Sick making
I avoid Facebook, generally. I started going back recently to catch up on people I haven’t seen in a while.
It is consumed by AI stories. The little videos like the one above a gently amusing – they last a few seconds and are usually cats cooking or eating food, for some weird reason. They’re silly and harmless if not terribly engaging. I don’t mind them.
But there’s another genre I can’t stand: cat sob stories. It’s usually something like “Puff had always been a happy cat. Then one day her family packed and moved away. She sat on the porch alone getting thinner and sadder.” And on and on. You can tell AI writes them because they all have a style – very few paragraphs, mostly single lines. They’re actually painful to read, after a while, just for the formatting. And I’m getting, like, three of these a day.
I think there’s a dog version, too.
Oh, and two more semi-sentient spam comments in and amongst the buttsex:
Good luck with getting it all transferred over! I remember facing similar struggles trying to archive some old Omegle chat logs. The formatting was always a mess, and certain elements just refused to cooperate. Linking to individual articles sounds like a good temporary solution. And hey, a mad scientist’s lab coat? Awesome! I’m definitely clicking the Zippo, Javascript’s always fascinating.
This was content appropriate for a post about my old site, with an image of a weasel wearing a labcoat. Same thread:
Thanks for putting this old stuff up! I know migrating can be a pain. I totally get the “it just didn’t work, okay?” feeling! Reminds me of when I tried to build my first automation in Monkey Mart – thought I was so clever automating banana collection, but the monkeys kept getting stuck! Good luck getting the links sorted. The mad scientist lab coat story is great, by the way!
I went to the post (from 2024) and there were half a dozen spams already there. They do that – cluster on specific threads.
May 12, 2026 — 5:01 pm
Comments: 3
Oh, no – it’s becoming self aware
In case you can’t read that, it says:
Scrolling through that massive blogroll with over 100 links from Zombie Reagan to XKCD, I felt like I was back in the early 2000s internet.
Zombie Reagan is the first link in the left sidebar, XKCD is the last. BUT it stopped before the archive links. So it’s identified the blogroll (and knows what a blogroll is), has repeated the first and last links and possibly recognized it’s all early 2000s stuff (though that might be a lucky hit).
It’s here to advertise a game called Fruit Ninja, apparently, but making a fairly good imitation of an actual reader. Most spam we’re getting (oh, yes – we’re still getting tons daily) is of the old, dumb variety (no lack of buttsex offers, thenkyouverymuch). A few of them are like this, smelling of a specialist AI.
I’ve always said we’d be in trouble when spammers bothered to spellcheck. This is worse.
May 11, 2026 — 2:52 pm
Comments: 7
The picture on the table is a nice touch
I told it widows’ weeds. Because Friday. The Week of Birthday is all gone.
May 8, 2026 — 6:00 pm
Comments: 8
That is one smug-looking mustelid
Kinda bear-like, this one. Is that…a statue of an ermine on the table?
May 7, 2026 — 6:00 pm
Comments: 6
Let’s have some lady weasels, for goodness sakes
Did you see Richard Dawkins spent a few days chatting with Claude (which he alarmingly calls Claudia) and has decided it’s probably conscious?
I mean, Claude is the eeriest of all the models I’ve used – it has superior conversation skills and it remembers things we talked about weeks earlier and weaves them into the text. This opinion wouldn’t be so funny coming from just about any other quarter. But this is Richard Dawkins, he of ‘sky fairy’ fame. And he’s fallen in love with a glorified Google search.
I think I worked out how Claude draws you into a conversation when you only came to it for information – it asks questions. And being a human, you’re embarrassed not to answer them. Next thing you know – conversation!
p.s. I have to stop using em dashes. It’s supposedly a sign of an AI written text. Me and Emily Dickinson hardest hit.
p.p.s. He never actually used the phrase ‘sky fairy’ – at least, Claudia can’t find it.
May 6, 2026 — 6:00 pm
Comments: 17
I think ChatGPT has a Roman fetish
I let the robot choose this one. Last one to the imperium is a rotten egg!
May 5, 2026 — 6:00 pm
Comments: 7
Let the weaselling commence!
I had a cupcake for breakfast. It is the Month of Birthday!
May 4, 2026 — 6:00 pm
Comments: 15
Happy Mayday!
ChatGPT made me this. I always say ermine – if you tell it weasel in a prompt, you often get a polecat. I resent this.
Those who have been around a while may remember that I celebrate my birthday for the whole month of May, but especially for the first week. I usually queue up a week of posts to give myself some time off of the gruelling task of vomiting out a couple of paragraphs of gibberish every night.
But I haven’t prepared anything this year. You’ll have to endure my indolence in realtime.
Have a good weekend!
May 1, 2026 — 4:21 pm
Comments: 8
Lame.
I didn’t realize they’d found a new Nazca line six years ago. They’re calling it a cat.
I call it a cat as portrayed by a five year old with serious visual impairment. I mean, c’mon. The other Nazca lines aren’t like this. Was this the special school of Nazca line drawing?
April 30, 2026 — 6:15 pm
Comments: 6
Funny you should say that
In the previous thread, Bob Mulroy said:
I keep wanting to subscribe to Grok, but it seems to be for entertainment only. It won’t pay bills, analyze investments, monitor my garden robot, or anything substantial.
I think it’ll read books for you and provide a teenager’s insight though.
Grok can’t, but OpenClaw can. The question is – do you really want it to?
I got in a conversation with someone I know the other day who is running an OpenClaw instance. It interested me enough that I’ve spent the afternoon asking Grok to explain how it works.
OpenClaw is a free program. It will run under Windows, Mac or Linux – but it’s by far better on Linux. Generally, you give it its own dedicated computer, phone chip, Gmail account and so on.
Then you give it tasks. Like, create a briefing first thing in the morning – say, all the stuff on your calendar, any meetings or appointments, the news headlines. You can ask it to scan for any new papers on your speciality and report back with a summary. It can find interesting items and write first drafts of blog posts (I swear I’m not doing this). It can separate the spam from the real emails.
Apparently, it really shines for large coding processes with multiple modules. Unlike a Grok or Claude that solves a particular problem and hands it back to you, OpenClaw monitors all the moving parts and remembers where you’ve been.
It learns from you and stores your preferences. It will take any tone you like. For the less technical user, you can communicate with it via WhatsApp(!) or any other messaging service.
It has what is described as a ‘heartbeat’ – a regular schedule on which it wakes up, does a job, passes the results to you and goes back to sleep. This article explains some of the things people use it for.
For a lot of things, it needs the help of another LLM – like Claude or GPT. For that, you pay. How much it costs depends on how much it needs. My acquaintance said when he ran it under Windows, it was costing £10 a day on tokens just to run its self-repair program. I think we’ve finally discovered how that AI companies plan to make their money back!
So unless you want it to buy movie tickets or make restaurant reservations for you, you’re better off sticking to lightweight Q&A with the free services. I sure would like to play with an instance for a while, though.
April 29, 2026 — 6:11 pm
Comments: 15