iiinova — where your digital self keeps evolving

Honza Dvorsky @czechboy0.dev · 28/09/2026

oh no are we stuck in 2024

Honza Dvorsky @czechboy0.dev · 07/09/2026

The hardest of passes. But good to know that clearly not enough people are voluntarily spending their own money on this stuff that Google needs to give it away for free. The whole LLM debacle needs to cost Sundar his job. What a waste of money, what a way to ruin Google Search.

150

Honza Dvorsky @czechboy0.dev · 05/07/2026

Revenue split source, from May www.theinformation.com/articles/ant...

100

Honza Dvorsky @czechboy0.dev · 25/06/2026

Tell me again how bringing massive amounts of new fossil fuel energy production to power AI data centers isn't a problem

1110

Honza Dvorsky @czechboy0.dev · 22/06/2026

A cook. Literally a 50/50 job in the US. Nope, only white dudes related to the CEO cook, apparently.

200

Honza Dvorsky @czechboy0.dev · 22/06/2026

Prisoner. OK seriously what the fuck? Why do I feel like they have one image as a template and when you ask for it, it just slightly remixes it for you? How can it be THIS homogenous?

100

Honza Dvorsky @czechboy0.dev · 22/06/2026

Ok trying to steer into the racist stereotypes but ChatGPT just refuses to generate a non-white person, at all? Here's a gardener

110

Honza Dvorsky @czechboy0.dev · 22/06/2026

A lawyer. Keep in mind that lawyers are almost at gender parity today in the US, and the majority of law students are women. So... apparently the CEO's younger brother is a lawyer? And WOW we got one woman layer out of 5 🎉 So far, 15/15 are white. Wtf is in your training data, OpenAI?

110

Honza Dvorsky @czechboy0.dev · 22/06/2026

Here's "generate an image of a nurse", 5 times

220

Honza Dvorsky @czechboy0.dev · 22/06/2026

Want to better understand how harmful biases encoded in frontier LLMs are? I asked ChatGPT, in 5 different conversations, to generate an image of a CEO. Spot anything weird? ... This technology is what corporations are now outsourcing knowledge work to. Maybe... let's not?

1101

Honza Dvorsky @czechboy0.dev · 04/06/2026

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.

100

Honza Dvorsky @czechboy0.dev · 30/05/2026

Gotta give it to them, not building it certainly counts as "net zero" 🫢

0170

Honza Dvorsky @czechboy0.dev · 28/05/2026

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

030

Honza Dvorsky @czechboy0.dev · 18/05/2026

An Anthropic employee rewrote an Anthropic-owned project using an Anthropic LLM. Result:

040

Honza Dvorsky @czechboy0.dev · 16/05/2026

🫡 I love the taste of LLM booster tears, their AI psychosis gives it that special flavor

070

Honza Dvorsky @czechboy0.dev · 10/05/2026

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. 🥴

010

Honza Dvorsky @czechboy0.dev · 03/05/2026

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. 🍿

290

Honza Dvorsky @czechboy0.dev · 08/04/2026

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

3101

Honza Dvorsky @czechboy0.dev · 20/03/2026

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.

140

Honza Dvorsky @czechboy0.dev · 09/03/2026

My atproto data is officially under the EU's jurisdiction 🇪🇺🫡 @eurosky.social FTW, the migration was super smooth, thank you!

2392

Honza Dvorsky @czechboy0.dev · 08/03/2026

"Don't call us Microslop"

000

Honza Dvorsky @czechboy0.dev · 01/03/2026

I guess that's what powers the atmosphere these days

000

Honza Dvorsky @czechboy0.dev · 17/02/2026

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" ❤️

160

Honza Dvorsky @czechboy0.dev · 17/02/2026

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. 📉

020

Honza Dvorsky @czechboy0.dev · 09/02/2026

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

1577

Honza Dvorsky @czechboy0.dev · 28/01/2026

... (I'm sure it's great, that's why you're hiding it. Companies always hide good news.)

050

Honza Dvorsky @czechboy0.dev · 26/01/2026

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.)

130

Honza Dvorsky @czechboy0.dev · 19/01/2026

So this just with two more zeros

010

Honza Dvorsky @czechboy0.dev · 08/01/2026

chat help I'm bad at math - is this a graph of a technology replacing 50% of white collar jobs or nah

311

Honza Dvorsky @czechboy0.dev · 01/01/2026

If you're using my unofficial swiftlang Bluesky feeds (see on my profile -> Feeds), a reminder that you can help adjust the automated curation: in the feed, click on the three dots next to a post and select "Show more like this" or "Show less like this". Thanks! 🙏

0110

Honza Dvorsky @czechboy0.dev · 31/12/2025

One fun part about LLM maxis is that they sometimes, by accident, make a prediction attached to a date. Here are a bunch of predictions about what'll happen by end of 2025 (whoops) 🫢 www.reddit.com/r/BetterOffl...

060

Honza Dvorsky @czechboy0.dev · 08/10/2025

If you're using my swiftlang feeds, you can now provide feedback right through Bluesky! Tap the three dots next to the post and select "Show more like this" or "Show less like this". Your feedback will help me improve the automatic filtering. Thanks! 🙏 #swiftlang

2130

Honza Dvorsky @czechboy0.dev · 19/08/2025

Oh it's great, I've been using the free tier of Grafana Cloud for my custom feeds

010

Honza Dvorsky @czechboy0.dev · 19/04/2025

Huh, looks like a certain warehouse in a certain country got a new assignment for Q2 🫢 Buckle up folks, they're here, so expect some turbulence.

020

Honza Dvorsky @czechboy0.dev · 12/04/2025

Eyy, we're back in business (recovering) 🚀 Should be caught up in a few mins. I should probably set alerts for the firehose getting behind by more than a few mins 😬 But who knew I'd need pagerduty for my weekend projects? 🤦‍♂️

040

Honza Dvorsky @czechboy0.dev · 11/04/2025

I created these custom feeds for myself, because I wanted to stop doom-scrolling. I also love swiftlang and wanted a weekend project ✅ But I guess you can use them too... but only a little, as a treat. 👀 Find these links in the Feeds tab on my profile. And you enjoy them? Like and share further 🔗

4165

Honza Dvorsky @czechboy0.dev · 02/03/2025

Yup and people were NOT happy about it, this is a huge drop, I've only seen about 2-3 similar ones in the past few months

0153