Boston-area meat construct ␥ I just do what the plants tell me ␥ I'd rather be undermining the client-server paradigm
This is the more tech-y alt of https://cybersecurity.theater/@varx
| pronouns | he/they |
| languages | 📖 en, es; ✍️ en, ~es |
| that cavern thing I'm always nattering about | https://codeberg.org/cavern/docs |
RE: https://neuromatch.social/@jonny/116666900898570791
this on its own is such a spectacularly bad idea it effectively renders the project dead
I think the modal situation here is that the people are reading none or very little of what is being generated by the LLM, so the tests have a special role: Tests function as the pull arm on the slot machine, you just generate until tests pass, and that's a jackpot. Obviously that's meaningless when the tests are meaningless, so tests take on a very different meaning and role in slot machine coding.
Previously we would write careful test conditions that were based off some real problem or an understanding of what the code under test did, and had a specific thing they were intended to protect against. Tests move slow and are designed to protect us against the things we know can go wrong. When we learn of a new wrong thing, we add a test.
LLM tests have the form of tests but don't do the same thing. They often test nothing, and are just expressions of truisms that the probabilistic text space explored while generating. They have strongly worded names but end up actually asserting that basic language features work as expected. Because it is not us writing tests for ourselves, where we only harm ourselves by making them weak, they function instead as a passively obfuscated justification for the code that the LLM generates. The user wants the tests to pass. The LLM provides.
The tests are theater: they are the play field for the slot machine. They are mild, surmountable, need to fail a few times to be plausible, but must eventually pass within the expected generation loop window to deliver the payout.
new post, in which I outline the absolute absurdities of recent 0-day disclosure, and curse myself for having any foresight on what might happen in the coming years.
read it here: https://sdomi.pl/weblog/29-please-do-better-thanks/
(and do bring the tea…)
We're hiring at my workplace.
If you are interested in working
- in a science-adjacent nonprofit
- in #python
- doing web backend and data engineering stuff
- *not using generative AI*
- remote work friendly
Please drop me a line! Your application won't skip the queue but I can give you a boost.
I rarely get a chance, since we're so small, but would love to help someone on here #GetFediHired !
Please feel free to boost for reach, or forward to your friends!
How are y'all handling coworkers who post slop?
Several of our contractors have made rather voluminous wiki pages that are heavily redundant and over-explanatory. So far my approach has been to just... quietly not read them, and pretend the pages don't exist. (If I need information that the page is supposed to have, I just ask the contractor to explain in Slack or a meeting.) It's bad for the company in a bunch of different ways, but the company is all-in on AI and doesn't want to hear dissent, so there's no way to address this systemically. (And I'm not invested in the company's long-term health.)
One coworker posts AI outputs sometimes, but is a bit more discerning, and we have a good enough relationship that I've been able to explain that hey, I'm not reading that, but you're free to tell me anything you learned *after* you verify it.
I'm curious to hear how others are handling it.
RE: https://hachyderm.io/@inthehands/116607455389757472
I never thought this day would come, but this is a pretty stupid future, so...
I've added noindex,nofollow meta tags for Google user-agents in all of my site's page headers.
RE: https://tldr.nettime.org/@tante/116605858023186072
Google Search rests on a social contract: their bots can crawl our sites, they can index our sites, and they can show excerpts of our sites because
and •only because•
they send people to our sites. •Our• sites, our words, with our design, with our links, with our context and our aesthetics, shared the way we want to share them.
Google is announcing — unambiguously and with great fanfare — that they are now fully breaking that already-ragged contract. We should reciprocate.
1/2
I always find it funny when someone emails security@ after having performed some deep research against the site and discovered a critical vulnerability... and then asks what our disclosure and reward policy is.
Which is linked from the footer of the main page.
Here’s how AI would do your laundry if it worked the same way it currently writes code.
1. Put all laundry in the washer.
2. Door won’t close because it’s too full. Go to the hardware store and buy some scrap metal, rubber, hinges, screws, bolts. Build a bigger door and attach it so all clothes fit.
3. Water now pools in the bottom of the new door during wash and doesn’t fully drain. Cut a hole in the door and weld a Shop-Vac hose to it to suck out the water during spin dry.
4. There isn’t a free outlet for the Shop-Vac. Install a new electrical box with more outlets.
5. Too much power is now being drawn and the circuit breaks. Buy a generator and couple it to the existing electrical.
6. Neighbors complain that the generator is too loud. Murder the neighbors.
7. Generator runs out of gasoline. Dig arbitrarily around the yard for oil veins and write a Markdown document for building a refinery.
[Maximum tokens exceeded]
jcoglan
jonny (nonvenomous)
dmi 💽 soon --> GPN
Stephen Rosen
Paul Cantrell
Jay 🆘