upstream source or tl;dr for Youtube boycotters?

The author's relevant experience is in developing Copilot, not in information theory or statistics. So, when they say:

Unlike image and video watermarks, text watermarks will always be trivial to remove.

This is wrong; image and video watermarks are also trivial to destroy. In general, watermarking only works for physical objects and fails for information-theoretic objects.

Watermarks fail. PersonalIP survives.

suspect merican llms will go the way of the shipping industry. technically capable protected species that's globally irrelevant.

Curious what happens to Qwen with Anthropic calling out their distillation to the government.

Doesn't that just mean that many people in the internet wish that Trump dies of rabies?

To be fair. 🎶 To. Be. Fair. 🎶 DuckDuckGo also offers NoAIDuckDuckGo. NoAIDuckDuckGo

That would be fitting.

Can’t replicate.

Sucks when you're on the receiving end of it, huh?

sue their asses I assume? not that you can fight an army of lawyer sharks but anyways

By using such a formulaic reply, as an LLM would, I'm emphasising your point. 🤣

My point, and that of the author of the article is not to defend AI slop, rather it's to offer real criticism, instead of making the accusation that something is AI generated:

That isn’t criticism. It gives the author nothing to fix. No sentence. No claim. No result. Just a little accusation dropped in the middle of the room. Just sending bad vibes for no reason

And it is rude to expect people to proofread your slop.

The author isn't defending slop, in fact they hate it too:

I utterly hate AI-generated content. I don’t want machines to replace personal writing with synthetic filler.

What they are arguing for is AI assisted proof reading tools, particularly if English isn't your native tongue:

But a spell checker doesn’t make a post fake. Grammarly didn’t make old posts fake. A machine helping a non-native speaker write more readable English doesn’t make the work fake.

Nah, just having some fun.

By using such a formulaic reply, as an LLM would, I'm emphasising your point. 🤣

If OP doesn't recognise it, that's probably to their credit!

They are responding to you as an llm would

Would you like to share any other social insights to continue this line of thought or would you like to find other places to share this thought so others can benefit from it?

Could you please explain what you're trying to say? Are you saying that a programming community on the topic of AI, is not an appropriate place to share this article?

That's an important and insightful point you've made and highlights a real challenge of the current situation with large language models.

Would you like to share any other social insights to continue this line of thought or would you like to find other places to share this thought so others can benefit from it?

Ahh yes, who could have seen Meta having issues with privacy...

And absolutely exciting to me! Because it's so counter the "my daddy has the bigger parameter set" the corporate kids play.

There are however a number of people that have attempted to do LoRA with tool calling. It hurts a bit of the performance in reasoning and it makes it over eager to use tools, but this is being tinkered with

It's an interesting experiment but note that it'soptimized for these benchmarks.

They describe it quite clear themselves in a warning:

🚨 1.This model was not trained on tool-calling or agent-based programming data. We therefore do not recommend using it for tasks that involve function calling, API orchestration, or autonomous coding agents. For programming tasks, we recommend using this model on competitive programming problems (e.g., LeetCode-style).

Can't stand 'an event has occurred!' vagueposting headlines, so I searched GLM 5.2 and found their HuggingFace overview. Wondered if this was another model getting good by getting small - then had a laugh at the files starting "model-00001-of-00282.safetensors (5.34 GB)." So uh. No.

I'm not caught up on all the details, but as I said this has been an ongoing game of cat and mouse with people looking to leverage other models to improve their own (this isn't just a qwen vs Claude thing and everyone if pointing fingers at everyone). If I had to guess it's only been more recently that AI companies are actively fighting these issues, but even then I imagine they're not 100% effective.