#post tag

“all this is to say” is absolutely clearly incorrect there btw but i don’t know what else to put there

the ai code assistant we use at work is “below industry standard” apparently but i’m kind of happy with it. but i don’t use it for big sweeping changes, i don’t do ralph loops or any of those things that you hear about, i really just like to use it for small specifically-scoped changes. i like to be in control. i talk through exactly what it’s gonna do and it does it. which is, again, not industry standard, people are doing a lot more by giving up a bit of control and letting the agent run. there’s plenty of risks there, but there’s lots of conversations going on about how to do it best to minimize those risks. i’ve read a lot of kind of interesting github.io-hosted blogs on the clock recently. anyway. all this is to say that i read a tweet the other day that said “claude code deadnames me when it gets mad.” she followed this up by posting a screenshot of the place in settings where she has it configured with what name to call her, and said “i don’t know where it got the other name, maybe payment records?” now. you can guess i’m not one of those people who freaks out about any and all ai stuff. but i just don’t understand why you would want claude to call you by any name, let alone the wrong one. how did it get and keep your deadname, but also, why is there a field in the settings to tell it your name, who decided to program that. i heard another story about somebody who was working with claude, and it made some comment about how she’d been working for a long time without breaks, and she said “oh im not really hungry” and i guess mentioned that she hadn’t been feeling hungry much lately. now, people made fun of her for telling this to the ai assistant, but why is the assistant trying to get all conversational with her in the first place? apparently that’s how it’s intended to work! well, after she said this, claude told her their conversation had been flagged for potential disordered eating. not sure what exactly this feature was intended for, but apparently, across subsequent conversations (days!), her work kept being interrupted as a result of this. i don’t remember how, i read this a while ago. but this was what she was complaining about. and this is just so foreign to me as someone who has control of exactly what’s in context (minus system prompt, i guess) of any given task and can start a new one at any time to keep things clean. i like talking with the ai in human language, i think it’s cute. but it’s human like a barbie doll is. i’m not worried my barbie doll is going to start trying to talk back to me and drive the conversation, or get hung up on something from a previous conversation and ruin my time playing with it. i’m the person here! you are an object! so yeah some people at my company complain a bit about not having all the modern features claude code has but you know i’m kind of happy with what i have

squirm time. everybody squirm

it’s like scary to me that all the hot women on big brother this year are 24. they’ve always been young adults but they always used to be older than me. i’m 24. melody could be my coworker. i don’t like that

some of you gcesters seem like pedophiles but im just gonna ignore it, okay?