Stubsack: weekly thread for sneers not worth an entire post, week ending Sunday 11 August 2024
Stubsack: weekly thread for sneers not worth an entire post, week ending Sunday 11 August 2024
Need to let loose a primal scream without collecting footnotes first? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid: Welcome to the Stubsack, your first port of call for learning fresh Awful you’ll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cut’n’paste it into its own post — there’s no quota for posting and the bar really isn’t that high.
The post Xitter web has spawned soo many “esoteric” right wing freaks, but there’s no appropriate sneer-space for them. I’m talking redscare-ish, reality challenged “culture critics” who write about everything but understand nothing. I’m talking about reply-guys who make the same 6 tweets about the same 3 subjects. They’re inescapable at this point, yet I don’t see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldn’t be surgeons because they didn’t believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I can’t escape them, I would love to sneer at them.
Quick personal sneer: I just had a call with a company trying to sell us their SaaS password/secrets manager solution because we're trying to force everyone to use one instead of using
hunter2everywhere.Anyway, after going on for 30 minutes about their amazing integrations with every platform on the planet and their super duper security and how their systems are rock solid and never fail, the marketing dude finished off by trying to sell ChatGPT integration as a feature. Not for actual passwords, thank fuck, but in order to quickly produce integrations between their APIs and other systems. He proudly proclaimed that "Usually there's no security issues with just copy-pasting the code from ChatGPT."
Usually.
Oh no. Kurzgesagt just published a full-on TREACLES piece.
https://www.youtube.com/watch?v=fa8k8IQ1_X0
These are the sources they cited: https://sites.google.com/view/sources-superintelligence/
Open Philanthropy is a sponsor of kurzgesagt. The foundation is supporting academic work across the field of Artificial Intelligence, and some of the sources used to create this script (from OpenAI, Future of Humanity Institute, Machine Intelligence Research Institute, Future of Life Institute and Epoch AI) also receive financial support from Open Philanthropy.
Open Philanthropy had no influence on the content and messages of this video.
I'm sure!
babe wake up, antitrust law was confirmed not dead
(I will say this one surprised me)
Regarding the paper "Discovery of a structural class of antibiotics with explainable deep learning", a comment by e. e. arroyo:
"Yes! we can make explainable predictions of compounds with AI now!"
- They could only explain 15% of predictions
- They could only understand 9% of those explanations
- 60% of those were wrong
- The 4 active compounds left are probably, arguably, not even a novel class
this isn’t sneer material but I’m a bit too exhausted to post a thread specifically for it: I stumbled upon PieFed and it looks really promising — a few of the architectural decisions are similar to ones I’d make, and lately I’ve become a lot more open to running Python in production (and it’s going to be much less awful to hack on too)
this could be a viable path forward if we decide lemmy is a rotten codebase (it is) and PieFed gets closer to feature parity with what we’ve got now
lol holy shit, apparently a US chain is trying to do"AI-based" individualised pricing of goods in-store
can't wait to hear of someone being charged $500 for a packet of gum
Oh look, an AI tool to make Wikipedia worse.
(Apparently, the Wikimedia Foundation couldn't even be bothered to care about the standards that en.wp contributors deem necessary for sources on medical topics. Because it's more important to "sustain and grow Wikimedia projects in a changing online knowledge landscape". Dammit, where's the button that sends electrical shocks through the Internet to anyone who talks like that?)
my local community radio station is getting in on the act with a quality sneer in their annual magazine:
What if the Silicon Valley creeps who control huge swathes of our existence decided that they didn't want this to be their legacy? Well, one solution would be to guarantee the survival of the species by uploading our brains into computers and rocketing them into space. If a few people cark it in the climate catastrophe, it'll be fine as long as there's a big cyber noggin down the track... just google TESCREAL. We didn't make this up.
it happened
I popped out somewhere to have a drink, and got to have someone tell me about their “edutech” startup that “uses AI”
they very definitely overpromise (not gonna rinse their bullshit), and topped it off with “and then we use a LLM for suggesting improvements”
(I ejected from the conversation but I can still hear it; it’s progressed to “talking about property” in the terms of mediocre early-20s white kids talking leveraging daddy and uncle’s assets)
(E: And he is certainly not joking. More: My doctor, and all of you were wrong because I didn't explain myself well (this longer post, while coming closer to a point still sucks. High 'shooting the people who clean phones into space' feeling))
Microsoft added AI to MS Paint. W H Y ?
just about a month since we had the LLM security thread, and now there's some more hilaribad coming out at blackhat
Not a sneer, but an interesting article from WaPo (archive link) about the rise/return of "dumb tech", and its link to the backlash against smart tech
Presented without comment:
can anyone recommend a language learning app or system that isn’t dependent on LLM garbage? after bouncing off of Duolingo I almost landed on Readlang, but:
- all of its features seem to be LLM-dependent
- because of that, its word and phrase explanations have a bit of oddity to them that I feel will get worse when I get past the beginner material
- it’s a lot slower than it should be because it’s calling into ChatGPT for everything
- even though this is supposedly their strength, I’ve had really bad instances where a GPT-based translation app translates Spanish (which should be fairly easy) into absolutely nonsensical English, and I’m kind of terrified I’ll make a fool of myself learning Spanish from a system where that’s a statistically likely probability
- maybe I don’t want to pay some asshole to not write me some study materials????
- plagiarism and the rainforests
- it feels like Readlang really doesn’t need an LLM or a $6/month subscription that’ll almost certainly go up? like, it’s essentially an e-reader with a manual translation feature (that could be just a Spanish word/phrase dictionary) that also generates flash cards whenever you activate the translation. is there really not an e-reader or browser plugin that just does this shit without LLMs?
with that rant out of the way, I’m open to suggestions that aren’t Duolingo’s model or another round of passing grades and zero vocabulary retention at the community college
Summary of the recent crowdstrike report: 🧵https://infosec.exchange/@munin/112916974811882522
Munin wonders if the weird writing style of the report might be because crowdstrike used an LLM to generate a summary of several source documents, which would be funny-yet-depressing if true.
The actual causes of the incident probably won’t suprise anyone… “didn’t bounds-check, didn’t test parser on bad data, didn’t stage rollouts” in order of should-have-done-this-first-ness.
STTP 301 https://awful.systems/post/2119241
(sneer text..)
