A few years back, I wrote a post titled Large Language Models in 2023; I’ve been looking to write a sequel. I don’t know why. I’ve spilled a lot of (digital) ink about it in recent months, to the point where it feels less like “The Weekly Miscellaneous”, and more like “The Weekly Vent About AI”. Still, I want to write a post that concludes the series, that summarises the feelings I’ve had about the whole thing, why I’ve been writing about it so much and what I aim to get out of all this venting.

In writing this, I unearth a lot of hurt and anger — well, it was there anyway, now it is on a page. This could very well resonate with a lot of people. This could very well make people uncomfortable, even potential employers whose patronage I need to survive. So, the post gets drafted and re-drafted, as I attempt to find a suitably nuanced angle on an already oversaturated topic, and find this nuance despite the hurt and anger I feel.

Finally! A post that isn’t about sodding LLMs!

I went to EMF over the weekend, and fully expected it to be one of those camping trips where I’d be miserable for a lot of it due to sensory issues, social conflict, or some combination of the two, similar to how I used to experience Scouting camps. I was pleasantly surprised to find out the camp was not like that at all! (But it does make me wonder if the miserable experiences at Scouting camps could have been made avoidable….). I mostly used the time in workshops, learning practical and transferable skills, but I also met a lot of cool people, attended the Null Sector raves nearly every night, and visited a few talks as well. So, it ended up being a great experience and one I aim to take with me long after the camp.

As you may have heard, I’m putting together a list of software that has banned AI. I think it’s the first time ever a post like this escaped Fediverse containment, and I’m heartened to see so many people supporting my project and offering suggestions. I’m also heartened to see just how many projects have taken a firm and principled stance against Large Language Models (LLMs).

In compiling this list, I have read a fair few debates within software projects on whether a strict no-AI policy is suitable. Examples of this include the debate in Blender and the debate in GNU, both of which I read through in response to user suggestions.

I enjoy reading debates like this, and importantly I will note that none of these projects are attempting to defend vibe coding. Instead, they ask whether a policy banning generative AI is too strict or unenforceable, and whether it would be better to adopt a policy advocating for careful usage over an outright ban.

Content warning: This is an incredibly raw piece of writing; it discusses the topics of ableism, in particular my own internalised ableism.

I need to think about something else. I just wish I could.

It’s one of those days again. They are annoyingly frequent, leaving me with the feeling that doing any amount of effort, even a simple trip to the shops, is more like a hike up Mount Everest. It’s like there’s an invisible mental wall between myself and starting stuff that leaves me feeling demoralised, exhausted and useless.

I describe them to a friend as “low spoons days”. As in “I can’t even plan a trip out—even if I need the socialisation—because thinking about it makes my brain hurt”. Plans are for people who don’t have to plan around low-spoons-days.

I’m kinda sick of the Fediverse right now.

There’s been a huge amount of Discourse over Cory Doctorow’s post “Six Years Of Pluralistic”. Cory defends his use of Ollama and criticises those who eschew the technology, Tante writes a rebuttal in “Acting ethically in an imperfect world”, then a few hours later Tante has to tell people to stop attacking Cory, because people on the internet are completely normal in this way.

And, let’s face it, nobody looks good. Not the swarm of people attacking Cory, and also not Cory responding to arguments with the same tired points about the ethics of his individual use-case.

I think the only person who came out of this particular shitshow looking okay is Tante, who explicitly told people not to attack Cory and even wrote up a conciliatory post later on. I am not sure how much I agree with this post, but it has certainly got me reflecting on my own patterns of thinking.