biztos

I just went and tried this with ChatGPT and one of my favorite cartoonists, Olaf Schwarzbach aka OL (yes, German humor exists):

https://ol-cartoon.de/

I prompted ChatGPT: Make a cartoon in the style of "Die Mütter vom Kollwitzplatz"

What it generated looks nothing like OL's work at all, so short of the lame attempt at copying his humor and the correct setting in Berlin, the only way you would ever confuse it for an original is...

...by the quite authentic "OL" signature ChatGPT rendered in the bottom right.

I saved the proof but am not posting it anywhere. You can probably generate the same thing yourself. Below is a rambling discussion with ChatGPT about it.

https://chatgpt.com/share/6ac4d10f-9a70-83ec-ac39-7f80c6cc63...

How big are the images? Is 160px 1%, 10%…?

How do you think they will verify this?

With AI?

Out of curiosity, why should sports betting via a bookie be legal in the US? Currently it is not.

Yes but that doesn't mean it's not the singularity!

If I get a response from the OpenAI API (or whatever), the fact that the text is embedded in a JSON object that is obviously an AI response, and even tells me which model was used and how many tokens it took, then I already have a machine-readable format clearly identifying the text as AI-generated. Whether the recipient chooses to keep that when they use the text is not my problem, right?

As the API provider, I would be very happy to consider myself compliant based on this. But I have a feeling it wouldn't fly in Brussels.

So if an AI-generated thing is cut in half, and only one half can be identified as being AI-generated while the other half can not, do we think that will be OK?

Leaving aside the fact that this is already the new Ubiquitous Cookie Consent[0], how do you arrive at metadata on text output satisfying this requirement?

[0]: I was just in the EU and got a chuckle out of the "AI disclaimer" coming at the end of every second advertisement, soon to be every advertisement.

Both: English messages sent to me, and I also see my friends do it in Spanish for messages sent to other people.

I was surprised WhatsApp didn't give me a "transcribe" button (as recently as yesterday). Seems like it'd be pretty easy to implement that.

The WhatsApp voice message is very common in Spain. I always find it jarring when I'm there, mostly because I'm not used to it, but also because it seems to defeat the purpose of the messaging app: I can't listen to your voice message on the bus, for instance.

It's so normalized there that I wonder whether this is a particularly Spanish habit. Maybe there are a bunch of countries where people routinely leave each other voice recordings on WhatsApp/Messenger/etc, and I just haven't been to them yet.

Basket? Fish? All I see is the model recursively running itself locally on an eye-pad, which for some reason beyond our understanding is obscuring the invisible fork.

FWIW I did a day of German-language software design with Claude and it was great, a bit strenuous of course but I also wonder if it might have found some bugs it would have overlooked in English.

If you're doing it for writing practice, one of the good/bad things is that it's quite typo-tolerant (in any language). So you can have the chat with less effort, at the risk of practicing bad habits.

I read it as asserting that unlike in Asia, people do this work willingly. Which I don't actually believe: if you couldn't find qualified people for this work in Cambodia, how are you going to find them in Africa?

That's a great idea.

I've been thinking about how to avoid having AI-isms creep into my own voice, because I see that way too much in my tech-world friends. It's natural enough: if you spend a lot of time talking to somebody, you usually end up mirroring some of their speech patterns.

One idea I had was to start doing my Claude sessions in German. (That happens to be my strongest second language, no other reason.) Thus any Claude-isms I pick up will presumably not wander into my English, which is what I use for most communications.

This would have the added benefits of getting some much-needed German typing practice, and preventing me from ever lazily copy-pasting AI text into an email or a README.

While it could of course be produced in Xinjiang without using "slave labor," the US government banned those tomatoes in 2021 because of that risk:

https://www.cbp.gov/newsroom/national-media-release/cbp-issu...

If Costco were circumventing the ban it'd be a pretty big deal. I couldn't google up any indications that they are, so on balance I'd say it's "possible" in the same way my winning the lottery is possible. Can't rule it out, but reasonable people should probably bet against it.

TIL: Xinjiang tomatoes are something like 15% of the global market!

"Publish Own Site, Syndicate Elsewhere"

https://indieweb.org/POSSE

(There were a dozen other, unrelated things ranking higher on Google.)

If you put a layer between the customer and the LLM, a la ChatGPT/Claude, are you a Model-as-a-Service business? I think you are not.

But then how thick does that layer need to be?

I think the real win from these models is enabling well-funded/profitable companies to not be beholden to the big closed-weight providers. But I like the idea of letting a hundred flowers bloom for under $20M/y each.

That's a fun idea! I wonder if it'd work if you just used something like an e-Ink screen and had it generate random license plates?

I suppose "blocking the view of surveillance cameras" is probably illegal though.

Haven’t read it yet but here it is:

https://stratechery.com/2026/whos-afraid-of-chinese-models/

In this new age of AI, all the devs are Daves.

That's a great blog/website! I always forget where it is because of the weird domain, but now I'm adding it to my RSS aggregator.

Marginal Revolution is nominally about economics, and it's not entirely not about AI, but AI is a fairly small part of what they cover:

https://marginalrevolution.com/

I follow a fair number of people who would, back in the day, have been classic bloggers, but are now doing Substack, and rarely even mention AI. Here's one I like enough that I actually pay for it:

https://walkingtheworld.substack.com/

Thanks for the explanation.

Safari on iPad lines these up almost perfectly - the second line is a tiny bit wider, I didn’t even notice it at first.

That example had a tone mark but no vowels, so I will try one with both. E&OE.

    |ดื่มน้ำอยู่|
    |abcdef|

[edit] These are even closer, but still imperfectly aligned on my iPad.

Interesting. Thai characters can also blow it out, I imagine because of the difficulty mapping glyphs to width:

https://biztos.com/hey/thai-mermaid-chart.png

To my surprise, Sublime Text gets it almost right:

https://biztos.com/hey/sublime-thai-mermaid.png

I tried finding a Thai monospace font and using that in the HTML but it was worse, probably didn't have the box drawing chars.

Still a fun tool and useful for lots of ASCII cases!

There is a quite popular poster on Twitter/X that I was following, and really enjoying his content. He did what you might call shallow dives into pretty interesting topics, several hundred words.

After about a month of reading his stuff I started to notice the Claudisms. And once you see them you can't unsee them, they start to pop up everywhere. I have no idea if he was using an LLM to "improve" his writing, or was "vibe-authoring" the whole thing.

I unfollowed, and now I have this creeping feeling that much of what I see in any given day on the Internets is actually AI slop, and I just haven't seen the signs yet.

Here in rural USA, we were paying $150 for very slow DSL, and now we're paying about $50 for quite fast Starlink.

In Asia I was paying $50 for very fast fiber, but that was in a major city; out at the farm you're on the mobile networks. So if I build a house out there and can do Starlink, I will do it.

Plus, there's the whole Starlink Roam thing: in California this summer, I see more and more vans with the little Starlink rectangle on top. "Work from Campsite" is pretty compelling, honestly.

I just went to Apple.com and specced out the top of the line 14" Macbook Pro, and the price is $9,849.00. Well over 10K with California sales tax.

I could absolutely justify having that machine for the actual work I do, and I'm not even doing any of the really hard AI things. If I were, I might prefer a Thelio Mega workstation from System76, which is $90,383.00 fully loaded.

Alas I can not afford a 10K personal computer right now, which is not the same thing.

Hey, that's awesome! Very nice work and a really good portfolio site.

How did you find the festival-booth experience?

For my own projects, I'm very happy with an outcome that is "not faster, but better" as a result of my use of generative AI.

I still hope this will be a shared goal in at least some tech companies long-term. But the headwinds are strong. "Not better, but faster" is starting to look like a job requirement.