Honza Dvorsky

Tick... tock... the bubble starts to pop🍿

move slow and repair things

🏗️What if your whole stack was built with Swift?
@telemetrydeck.com made it a reality, building an internet-scale, privacy-first analytics backend from the ground up. Read their story: www.swift.org/blog/buildin...

Ai boosters are misinformation peddlers who have been lying about this since 2023 when they lied and said gpt 4 tricked a Taskrabbit into solving a captcha (it didn’t do this btw). These lies have enriched other liars and killed people and they are responsible

Thinking is not compatible with the LLM industry, that's why they are trying to create a tool to outsource it

a young boy in a yellow shirt is holding a cup of coffee and making a funny face . ALT: a young boy in a yellow shirt is holding a cup of coffee and making a funny face .

Non-Americans in the chat

I wish we were creating AI in the timeline where we didn’t also have an ongoing global fascist revolution, with which the creation of AI is inextricably entangled, and it’s hard to square any use of that AI we actually got with being opposed to that fascism (it’s easy to choose to not see the link)

LLM-generated code is not copyrightable, according to the US Supreme Court.

So a purely LLM-generated file in an OSS project should not have a license header, and is free for anyone to take. No license applies.

Pretty wild. LLM-generated code is now a poison pill for important OSS projects.

I give you $1b per year.
You give me $1b per year.

Now we created two businesses, each doing $1 billion ARR (much money, such economy)... yet no money changes hands, no value created, no services provided.

This is the gist of the LLM bubble, and this will be the plot of The Big Short 2 in 2035 🙃

Labor market impacts of AI: A new measure and early evidence Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

Anthropic's research shows that programming is the single biggest use case of LLMs today. Yet the same research projects over 15% growth in the number of software engineers between now and 2034.

But yeah, "programming is dead" 🙃

www.anthropic.com/research/lab...

Exactly, quartering AFTER a fair trial is totally cool

🇪🇺

Oh so the ChatGPT shopping experience that never actually seemed to work for anybody anywhere is actually total bullshit and disappearing? How many times does OpenAI get to just fucking lie before the media wises up?

They have it doing war crimes now. But yeah, very cool that you're doing more commits per week

Sorry I'm not more open-minded about LLMs, it's just some fucking maniacs shoveled out a bunch of useless bloatware featuring that technology, did not give me any chance to opt out, reorganized the entire economy around it, zeroed out gains made by green energy, and made it impossible to buy RAM

VSCode disassembling generated WebAssembly code from a Swift breakpoint.

Great language support means nothing without great tooling. Watch Jonas Devlieghere show how LLDB brings first-class debugging to Swift on WebAssembly. fosdem.org/2026/schedul... #Wasm #FOSDEM

Yeah again I've seen a lot of "this thing has very little value", I haven't seen anyone claim "this thing literally can't emit a single line of code".

I haven't so far seen a single person claim LLMs have zero utility. The criticism I see is the type you described as okay.

My view is that LLMs aren't useful *enough* to justify destroying the environment, stealing everyone's data, burning a trillion dollars, and giving the worst people more power.

LLMs can be sometimes useful and Why's post can be stellar boot-licking, I don't see why these would be mutually exclusive.

I mostly think LLMs are mediocre and I don't understand people who gave up their career to prompt full time and shill for billionaires on socials.

I don't see a conflict here

Read Anthropic CEO’s Memo Attacking OpenAI’s ‘Mendacious’ Pentagon Announcement Anthropic CEO Dario Amodei sent a 1,600-wordmemo to employees Friday as OpenAI announced a deal to provide AI to the Pentagon. The OpenAI move came hours after the Pentagon said it would sever ties wi...

Credit where credit is due, this internal memo from Dario is fun to read 🍿 (pay wall) www.theinformation.com/articles/rea...

I think more democratic governments should tell Trump to just fuck off instead of trying to be nice to him to avoid making him angry

The LLM industry is in a symbiotic relationship with Trump. Two grifts propping each other up.

Curious how the many employees at OpenAI, Anthropic, and Google square that circle.

Imagine coming to work every day to work on LLMs and knowing the biggest beneficiary of your labor is Trump.

I would never call it Microslop! That’s deeply disrespectful to Microslop employees and leadership.

A screenshot of the Polymarket website showing a prediction market titled "Nuclear weapon detonation by...?" under the Geopolitics and Ukraine categories. The market displays three betting options with their current probabilities and "Buy Yes" or "Buy No" prices: "March 31" at 5%, "June 30" at 12%, and "Before 2027" at 22%. The total trading volume shown is over $843,000.

Polymarket has created a betting market on the use of nuclear weapons. Everyone involved in this should be put in prison for life.

Every time I see someone suggest "we could use an LLM for this", I hear "I don't care about this being done at a high quality".

That's fine, just be honest about it.

When I see people suggesting using LLMs to answer support tickets it shows me they don't give a shit about their customers.

Powerful companies owned by sociopathic billionaires gave us mass surveillance and fully autonomous weapons in cooperation with an autocratic government?

Sounds like a setup to a classic James Bond movie.

Where's an attractive spy saving humanity when you need them. 🫢

We’re working on bringing free @swiftlangconf.bsky.social & #cloudnative workshops to AMS while we're there for #KubeCon 🌷 Join us on Mon 23 Mar–

[10:30-12:00] Building Cloud Services w/ Swift: A Hands-On Intro: s.apple.com/dr2F8e2FI7
[13:30-15:00] Container Install Workshop: s.apple.com/dN4e7D6zV3

This matches almost every person uncritically pushing LLMs I see.

Zero critical thinking, they just repeat words like "exponential" for things that are, provably, not actually exponential. 🫢

LLM boosting is a mass outing event of people who probably do more damage than good to their organization.

Only one way to find out