Personally, it seems to me we are going to need a grassroots genAI subsection in the anti-ai advocacy.

Re your edit, in the sugar example, most people know high fructose sugars are bad because of the advocacy against it and many give no thought to the other sugars.

For the people who think a little more nuanced about this, I take it as
. No BigTech AI
. No data stealing, surveillance-free
. No hyperscale mega data centres that spoil our energy and water reserves
. No patriarchal, (neo)colonial systems
. No algorithmic bias, systemic discrimination
. No intransparent trainibg data and opaque models
. No extractivist business practices, indentured labour or other practices of injustice involved

Yes good point, and like that visually GenAI is different so it doesn't use "AI" in two places, but some think that GenAI is not a really something that normal people understand whereas everyone knows AI. It is a tradeoff.

Edit: Think about "Sugar Free". Not all sugars are bad, but getting more specific loses your audience.

Now we'll all need to keep a lookout for the disappearance of these "AI" canaries...

Looking at it now, not sure the two "AI" is a problem. Think this is pretty good actually. Seems like the symbol could be used with and without the caption.

But the problem here is that if you're not very explicitly comprehensive, it looks like you're trying to weasel in acceptance of some specific kinds of "AI".

Right. It is hard to be comprehensive and also specific. Here is a version with just "AI Free". The issue here, and it may not be a problem is repeating "AI" in two places.

Interesting. Good contribution.

Generally agree. Trying to find the right words. "Human Made" and "LLM Free" have been suggested, but somehow seem clunky.

Why? Curious about your thinking here?

Good points. Thanks.

To a normie audience, the scope of "GenAI" isn't clear. They might think it just means slop that originated from a small prompt, not "paraphrasing", "grammar correction", "summarization", "style change", "visual filters", "resolution enhancement", etc. that are all technically "generative" (confabulating bullshit details out of the statistical models) but which normies don't necessarily think of as such.