Well part of the network of trust would be enabling a protocol or platform where individuals can control their identity and data, and thus use clients that don't use AI.

I'm also not sure I agree with the idea of echo chambers being bad or even a thing to avoid. You're already in various echo chambers of varying sizes, based on your interests, spoken languages, and so on. It would certainly be cool for people to just learn all languages and learn all about every culture and every point of view, etc., but that's simply not feasible. So who, then, decides which cultures and points of views to prioritize? Well I don't want the answer to be the some company or some nation, so it really has to be the individual.

Besides, a lot of the reasons an individual might have for not engaging with certain points of view can be quite reasonable. I don't want to force trans people to regularly expose themselves to posts by transphobes, for example. Society can handle that particular interaction not occurring. And once again it just comes down to who gets to decide which interactions are worth having and which aren't, and it's really going to come down to individuals and what they're individually comfortable with. Sure, a transphobe interacting with trans people might change their ways, but I think we can find better ways of fixing transphobes existing than building a platform where trans people become obligated to do exposure therapy for transphobes.

I think a lot of the "we have to avoid echo chambers" sentiments stems from an unfounded trust in liberal democracy and the free market of ideas. Time and time again, it's been shown that that just leads to allowing extremists to portray their ideas as having some level of legitimacy. It's what leads to fascism and hate (even though ironically you'll see people argue echo chambers do that instead).