Expand StopSlop from just AI-generated slop

fxgn

There's essentially no difference between AI generated slop and old fashioned SEO spam that has existed long before LLMs were a thing. I don't understand why StopSlop has to focus only on the first one and add the "AI" qualifier everywhere. I think all obviously spam results should be marked, not just AI

Users will have better search results

Pleonasm

What are the tentative predictive features that Kagi has identified to-date in the StopSlop initiative, which differentiate search results that are "AI Slop" versus "Not AI Slop"?

Fxgn may be right: it could be more useful to think about the problem as "Slop" versus "Not Slop." Additionally, from a practical model development perspective, solving the "Slop"/"Not Slop" problem might be less challenging than solving the "AI Slop"/"Not AI Slop" problem, if "AI Slop" and "Human Slop" share common variance.

k7absp2h

Definitely there are tons of spammy websites that aren't even made by AI. StopSlop is just looking at AI-generated stuff, but the problem is way bigger. You've got copy-paste spam, fake pages pretending other products (like fake landing pages or copies of DeepSeek and NanoBanana). There's also phishing and sites for fake tools that promise things that are impossible, like SHA256 decryptors. All this junk makes search worse, whether AI is used or not.