You mean like LLM filters? Right now it's all reputation based on IP and domain with a whole ecosystem of anti-spam companies like Spamhaus, SenderScore, ProofPoint, etc.
Using NLP / LLM spam filtering would presumably be either inaccurate or expensive or both. Someone would have to pay for it.
Bayesian filters are basically just a cheaper / worse version of what an LLM filter would do. Very easy to beat. Especially if the spammer is using an LLM to write a semi-unique email for each recipient.
What I'm trying to tell you is hat this has de facto worked for me during the past 20+ years. I get ca. 100 spam mails a day and they all get neatly sorted in the spam folder. There is no server-side filtering at all, my email provider allows users to switch that off entirely (and better should because it's very faulty).
As I've said, I'm not interested in theoretical arguments. All of my domains wildcard forward to the same email address, too. Filtering client-side has never been a problem.
That's pretty interesting if true. It must be because you're the only one still doing it. Using techniques just for your idiosyncratic inbox doesn't make sense to a spammer. But if every inbox was only doing this, then your experience would be much different. Nothing but unstoppable spam. Exposure = money. Why would they not spam you? They don't like money?
Using NLP / LLM spam filtering would presumably be either inaccurate or expensive or both. Someone would have to pay for it.