Honza Dvorsky
@czechboy0
Swift Server Engineer at Apple 🍎🇨🇿🇺🇸 he/him Check out and follow @swift.org This is my dev account, @honzadvorsky.com is my personal alt account. I only speak for myself. Signal: honzadvorsky.99
Tell me again how bringing massive amounts of new fossil fuel energy production to power AI data centers isn't a problem
2.1k people liked and agreed with the original wording, even if you don't - the community reaction is pretty clear, and this tone policing appears to just try to change the subject from the problem at hand.
Gotta give it to them, not building it certainly counts as "net zero" 🫢
I genuinely thought people were faking it. Humanity already invested a trillion into this shit, it's embarrassing and people promoting this garbage should feel bad
An Anthropic employee rewrote an Anthropic-owned project using an Anthropic LLM. Result:
🫡 I love the taste of LLM booster tears, their AI psychosis gives it that special flavor
Another fun one, "love LLMs" in bio, followed by various LLM boosters. All they do is defend Sam and Dario and attack anyone criticizing LLMs. Boy do I hope they're at least getting paid for this. Would be said to do this for free. 🥴
I sometimes get asked why I'm so confident that the LLM bubble is just a tiny subset of the tech industry forcing LLMs onto the rest of humanity, and that it will ultimately fail. This - this is why I'm confident. People absolutely HATE this garbage. Look at the reactions. 🍿
I'm genuinely interested in any LLM booster explaining to me how both Anthropic can have a model that's too smart that it's dangerous to release, yet they can't even reach two nines of availability on their services, which thousands of companies manage just fine. Pls explain
PSA: I blocked that annoying AI slop account from spamming the swiftlang custom feed. Please keep reporting whatever you dislike in the feed with the "Show less like this" option.
My atproto data is officially under the EU's jurisdiction 🇪🇺🫡 @eurosky.social FTW, the migration was super smooth, thank you!
The LLM trifecta, demonstrated by Microslop: 1. Steal someone else's work, no attribution 2. Launder it through an LLM, which makes it much, much worse 3. Ship it to production without a single human reviewing it I can't even be mad though, they gave us the gift of "continvoucly morging" ❤️
With all the hype around the "best coding model", Claude Opus, how much progress has Anthropic made in the past 5 months? Go ahead, guess. In the Agentic Coding benchmark, how much percent did it improve from 4.5 to 4.6? 20%? 50%? Exponential improvement, right? ... Nope. It got worse. 📉
I work hard on tracking the connections between the crypto and LLM industries, to show how it's all one big grift, and they just... put it on their website
... (I'm sure it's great, that's why you're hiding it. Companies always hide good news.)
Good news, everyone! The AI overlords need us to... *checks notes*... train the genius superintelligence to... *checks notes again*... understand YAML. And want to pay $150/hour to do it. Apparently these LLMs are genius coders, but YAML is too complicated for them. (I really hope it's a joke.)
chat help I'm bad at math - is this a graph of a technology replacing 50% of white collar jobs or nah