Benjamin Borowski 🍁

begs the question "by what/whose definition of good?"

3x on chad, 4x tribs, reds, wowza

got a seed?

Force of Habit, by Barker from the album Stochastic Drift

Fav track from my fav album of 2025

sambarker.bandcamp.com/track/force-...

It's now been 8 years since I deleted my Facebook account. I think it's one of the best decisions I ever made. I always hear people say "but marketplace" and it's notable how often our "musts" are related to shopping while being targeted. I'm starting to think my next big hurdle is deleting Slack.

Continuing my learning journey and have about 46 new sketches working towards my second album, but you can grab my first album for any amount of money here for #bandcampfriday #playbleton

rewritten in rust, right?

Ah, US only, 😭

Consider making a solid tech adoption by preordering my book about the cosmos, metaphors, and liberation
bsky.app/profile/chan...

whoa, love that cover! I'm on it 💸

Kinda wild that there's just a newly pervasive "ambiance of exploits" to consider now. Using computers at all becoming increasingly risky.

"Use rust."
"I haven't even started."
"Good. Start in rust."

(Experimental) Open Weight Models MiniMax M2.5 Beta Be cautious when working with untrusted data, as smaller models may be more susceptible to prompt injection.

I often try to get out of my head as a technologist and into the minds of the layperson:

"Hey, what is untrusted data?"
"Gee, what's prompt injection? Should I be worried about that?"
"Hang on, *more* susceptible? Does that mean the larger models are susceptible as well? Is AI not secure?"

The Ease of LLMs Making hard work hardly easy.

"Hyperfixating on optimizing your labour is also laborious, the sort of hypercritical and hypercyclical metawork that leads to burnout. It compounds because it feels productive, so we evaluate ourselves as getting more done, freeing us up to...?"

This is a basic and crucial point for any discussion of LLMs’ use in teaching research or writing. Whether it can ape us, fool us, or get facts right or wrong is, in the end, irrelevant. The LLM is not the thinker we are trying to encourage; the LLM is not the writer that we are trying to improve.

It’s a bit tiresome, but I think a good general practice when you feel your own institutional/professional life eroding under the forces of late capital is to focus on mentally connecting that feeling of struggle to the many others experiencing versions of the same.

I am starting to wonder if people who spend a lot of time talking to LLMs lose their stamina for actual critical analysis because they're so used to having their own priors reflected back at them in the guise of another's voice.

B.C. to end time changes, adopt year-round daylight time | CBC News March 8 will be the last time most British Columbians have to change their clocks, B.C. Premier David Eby announced Monday.

holy shit, we're doing it!

I admit I didn't read it completely. Yep! I think we're in agreement, no? What am I missing?

that's helpful. I didn't mean a lacking of general intelligence; they lack fire intelligence, what I would call applicative knowledge. do you know what you don’t know, and does that knowledge change your behavior?

hmm, I donno, I'm generally intelligent, but I don't have enough of whatever kind of intelligence i need to interpret this.

Yeah, this is what I was expressing. That as stakes rise, expertise also must, depending on the reliability of the tools you're utilizing. The doc you shared would probably get you to a point where you would damage the fire engine. Yet it's "correct" knowledge-wise. Knowledge, not intelligence. 😉

move Dreams to Planning Updating Dreams Dreams Delete Do you want to continue? Reject | Delete pages

I do a lot of blank staring at walls as a person who has been coerced into using and teaching agentic AI.

good stuff, the "space" as you say of...

expertise x reliability x stakes

is a really helpful expression. might snag that one for a post! cheers.

I find thinking about this with physical tasks helps. Ask 4.6 for a guide on operating the attached. Great, now I will plunk you down in front of it and ask you to supply an interior team who needs adequate water pressure to survive. 4.6 can totally guide you through it. You taking accountability?

Yeah, this is primarily why I moved away from "helping" online and primarily do real-time office hours these days. I wrote a little thread about the shame of feeling duped by AI posts when "helping". Reddit is particularly bad.

I keep telling Marie she should just show up in hockey gear.

AI-first design (AIX) is XP that preferences rare action over common action (UX). You can absolutely feel the intention in this type of design; it serves no one and costs everyone; is mostly coercive. The future of AI is largely interfaceless. The fact that it's becoming *more* visible says a lot.

I enjoyed it as well.

Apparently a bug I helped identify with QUIC that was making Notion unusable for me made its way all the way up to c-suite at Cloudflare. What can I say—the reverse midas touch is a rare gift. 💩 Delighted to have my mostly functional Notion xp back.

Combined with the increasing cost of hardware, this feels like an attempt to capture all forms of computation by users on to cloud computers; and not simply as a remote server, but every interaction the user could have with the computer is also intermediated through a corporate controlled agent