Can i add to this that i would actually REALLY love if there was an option under the AI serach to flag but show.
As we all enter this new paradigm, only having the any/exclude/only options is a limitation to me. Given the relatively 'expert' nature of the Kagi user pool, I think an option that is like 'any' but also visually identified (maybe boxed/different outline) high likelihood AI slop would help me. It goes to training users to be able to identify it for themselves as a second layer of evaluation and as a way of learning to trust/interpret the Kagi Ai slop model.
I think there are likely issues there in terms of that providing a potential data source to train against the AI slop model which sucks, but as a user I would appreciate that as a transparency measure.