Replying to
@Francehelder@mastodon.social the scary part is that you can never be sure if this is a joke or a serious post
Remote
Timothy Wolodzko
@tymwol@hachyderm.io
mastodon 4.7.3474 Followers
927 Following
50 Posts
Joined November 07, 2022
GitHub:
Replying to
@timnitGebru@dair-community.social it’s capitalism only when it’s you who gets the money
Replying to
@icing@chaos.social c'mon, if you do any rewrite, in any language, you would expect bugs and vulnerabilities. You always can stick with the old thing and don't touch it, so you don't break it, but "don't touch it" is not always a wise long-term strategy.
Replying to
@gabrielesvelto mentally some of the billionaires seem to be teenagers, so it could have same effect
Replying to
@zackwhittaker@mastodon.social It's not copyright infringement if done by AI.
It's not hacking if done by AI.
It's not a wrong answer (but probabilistic reasoning) if written by AI.
Replying to
@bagder yeah, and even if it does not make you merge 10x more code, it produces more slop code that needs to be scaned, fixed, and patched, so the demand on things like GH Actions is probably even larger than the "productivity" gain.
Replying to
@dgar@aus.social 🎶 keep rollin' rollin' rollin'! https://www.youtube.com/watch?v=RYnFIRc0k6E
Limp Bizkit - Rollin' (Air Raid Vehicle)
Replying to
@mustapipa to be more precise, this is 50 years difference in technology as they were using Nikon d5 camera, which is already 10 years old.
I know it's nothing new, but I'm still gonna say it: why do people use LLMs for coding? I have access to LLMs at work, so from time to time I test them (different models, different problems), but almost every time they return garbage code.
- Recently I asked it to write code and avoid a specific pattern that was introducing a security vulnerability. The LLM insisting on the pattern and re-introducing it multiple times, even if reminded that it should not.
- In other case, it suggested "performance improvements", on "you're wrong" it run the benchmarks and indeed, it was wrong.
- In other case, it suggested a "performance improvement" of caching a value instead of repeated function calls. When benchmarked it changed nothing, the compiler already seemed to figured it out itself.
- In many cases it introduced some useless local optimizations that only made code uglier and more complicated without any benefit.
and neither of the cases was some subtle edge cases, it was really simple stuff. It's basically a tool that copies and pastes the answers from StackOverflow until the code compiles and tests pass, with some ability to adapt them to local context, but that's basically it. Yes, you can solve many problems like this. Yes, this is what many people often did. But it has nothing to do with LLM having any clue what it is doing with your code, let's be clear on this.
Open-source enshittification is technically a diarrhea..?
There should be a 💩 emoji reaction on LinkedIn. It would totally boost engagement metrics.
Replying to
@maxleibman I’d rather ask who would pay the universal income. If AI takes your job and salary, I guess the AI company would pay you, right? They wouldn’t take the money and evade to tax heaven, right..?
Replying to
@grork but ain't this an industry standard already..? "This is our internal solution, but it is shit, nobody uses it. Install xyz instead."
Replying to
@sankao@mas.to @icing@chaos.social even if you don't add features code needs maintenance, and they didn't introduce hardly any new features. If you touch the code, you potentially introduce bugs. If you rewrite it, you'll likely create even more. The only question is how it would pay-out in the long term, the number of bugs in the new thing means nothing tbh.
Replying to
@guenther@chaos.social @sankao@mas.to @icing@chaos.social OpenAI and all the other tech companies already proven that they don't give a fuck about licenses, and that they can get away with it, so maybe let's stop pretending anyone cares about the open-source licenses..?
Replying to
@TechJournalist@mastodon.social so I guess, TL;DR is that they improved some of their APIs and made an "AI agents ready" fuss about it..?
Replying to
@gerrymcgovern Imagine a data center that would cause drought in Amsterdam, that would be something, right..?
Replying to
@rysiek A group of kids sells lemonade. One of the kids bullies the other lemonade stands, so that only them can sell the lemonade. The current logic was to let the bully sell the lemonade because otherwise we can’t get the lemonade from anyone. That way, the bully benefits.
If US chooses war to solve it, this is rather obvious consequence. I’m not picking sides here, just saying that in this situation this is not an irrational thing to do for them. They did a number of strange decisions, but this is not one of them. For military reasons, there’s simply no reason why US wouldn’t block it if it is already blocked for everyone but Iran. They kept it open probably only because Trump didn’t have balls to havoc economy even worse.
Replying to
@rysiek Sounds stupid, but keep in mind that so far US was doing nothing about Iranian commercial ships going out of the strait, so while everyone else was stuck, they were free to sail. That's even crazier than the full blockade.
Replying to
@twipped you could also add a bus factor calculation per file, e.g. like
for f in $(git ls-files); do
printf '%s\t%s\n' "$(git log --pretty=format:'%an' "$f" | { cat ; echo ; } | wc -l)" "$f"
done | sort -n
Replying to
@soph@grrl.me I worry about Ed. I worry that he'd be laughing so hard he could end up on ER from it...
Replying to
@flyingpenguin@infosec.exchange but they know that they are doing this because one day it could be them sending the distress signal, right..? Hope they know...
Can we agree that we already reached the peak of the hype cycle, or do we need people to still keep trying..?
Replying to
@jasongorman even if it was, let's say that due to AI you need to make your budget 10% higher for AI to work (optimistically), if you instead spend 10% more on improving things you already have, tech debt, trainings, or people in general, wouldn't it make you more productive as an organization?
I was listening to an interview with an American socialist. He was talking about a socialist agenda for the US and… he was basically describing Europe.
Replying to
@flyingpenguin@infosec.exchange "You should always verify AI's answers" is such a BS advice.
Imagine you had a dictionary that had a disclaimer "this dictionary can be incorrect, consult another dictionary to make sure we got it right", tell me again, why did I need the first dictionary..?
Either you can trust AI's answers, or you can't. There is no "sometimes it is wrong", because it literally means that every time you need to double check everything, making it useless.
- I will close the Strait of Hormuz!
- No, I will!
- I will!!!
- It’s closed by me right now!
- No, it is not, I am closing it!
…
https://www.nytimes.com/2026/04/27/us/trump-iran-proposal.html?unlocked_article_code=1.eVA.0vfu.000aVH8DhanP&smid=url-share
Why there’s no very complex numbers like -5 + (4 + 8i)j ?