How so? The categories of AI disclosure seem fine to me.

For that I've got a personal example. I have a professional project made in laravel a PHP framework. I made the choice to use it before the AI craze, I am maintaining the project but I didn't choose for the tech stack to embrace AI like it did.

The fact that these projects take long to make does bind you to the choices of someone else makes that you can't control.

But you did nail it in the head, it's the focus on personal choice that I'm aiming at.

For the nuance, I'm setting up a labeling system for the users to decide for themselves.

The second side of the comment about using libraries that coopt AI code. I think LLM use falls into a personal category. You can't control that someone down your tech stack used it but you can certainly not use LLMs.

If you find out that your tech stack is tainted but your workflow is human asking for a rewrite is unfeasible. I'd argue that the next project should use human-centric tools but that is something for later down the stack.

Second ps.: There might be a method of redemption for a project to move from tainted to handcrafted but it'd be easily verified. For example, Dave the diver has an LLM usage notice where they did the translations with AI. Should they re-do the translation work with actual translators (or let the community translate the work themselves) there might be a case towards bumping from tainted to hand-crafted.

I'll be honest, I feel fine tainted being the "standard" and handcrafted being a the selective seal of approval. I'll think things over. The chance for redemption creates subjective cases that give the people to say "why them but not me".

It's a conscious choice. I want to design the index to encourage people working towards the handcrafted stage. Any kind of AI usage immediately putting you on either slop or tainted drives the point home.

Thanks for the questions!

• The first question is the one that has the most nuance. I believe originally it'd simply be considered handcrafted but the introduction of a vibe coded emulator WOULD move it from handcrafted to tainted. I'm working on sufficient labels to be able to work out these nuances out.
• I said yes, skipping
• You still made the choice to use genAI, placeholder assets were a thing before and they still can be in the future. Errors happen with placeholder assets, specially those made to pass as real assets. Here we have precedent, Alters developers used GenAI and mistakenly (or not, we can't verify) left some AI assets in the release version. What tainted remarks is the usage of LLMs at all, doesn't matter the circunstance.
• LLM usage, tainted.

Ps.: I've added an extra phrase to the tainted description.

It is fair to judge a technology by the totality of outcomes it produces. You might find a subset of the technology that may be positive but you can't use a part of it to justify the whole.

It really reminded me of this image.

A partial perception of a technology might blind us to the overall net benefit.

Machine build to steal and plagiarize steals and plagiarizes. Shocked pikachu

And the existence of alternative uses doesn't take away the pain of someone who needs to sit on a bench and can't. The ringing noise that only teenagers hear on public parks because the council doesn't want them there. Alternative uses of a bad technology do not redeem a technology.

I'm not saying there aren't uses for it but it's basically a "I don't want to see poor people around" technology made manifest. Also, aren't there better versions? I remember that metros can have doorways that impede going onto the tracks without the need of spikes.

I personally beg to differ. You can think of things like hostile architecture, automated killing tools, etc.

The way we design tools is inherently biased. You wouldn't use a gun for cooking, you use it for it's intended purpose of possibly harming other beings.