Oddbean new post about login | logout Absolutely! I wouldn’t mind to have one for myself tbh. I am thinking that all notes from follows should be allowed, reposts from non-followed should be scanned at low scrutiny level, replies from non-followed should be scanned at different levels depending and npub reputation. All the parameters should be configurable by the users (levels per sentiment). This will require for the user to use a single aggregator relay, but I think it’s very much doable. 🐶🐾🤔
Would it be a possible add-on to the wine filter or could we have both running in parallel or interacting in some way?
They are technically redundant, so you could aggregate your other relays into one and read from that with “guardian” or doorman 🐶🐾🤔
THIS
On it! 🐶🐾🫡
Ladies, hold fast. Don't run off. This is all a bit complicated and it's a work in progress. nostr:note1jaqq53339lec3q7j6xsq28wfg4mshvflqmyva5gqu228whqamz3s8x9uha
Am I to believe, to condense this idea down to it's most basic elements, that you are suggesting an AI that will go out and mute people for you BEFORE you have some unsavory encounter? Could we go a step further and surmise that this could lead to the idea of 'pre-crime'? Somebody I follow says X and Y but not Z I simply cannot tolerate Z, and since people who speak of X and Y invariably follow it up with Z, I filter them out on the presumption that they WILL offend me in the future for something not yet come to pass?
Not exactly. No muting will happen, notes that user “tunes” to be not welcome, will not be delivered to the client. If user follows the other person, then they could have an option not to filter at all 🐶🐾🫡
I see. The possibilities of AI frighten me for the simple reason that I don't know what they are, and inversely, their limitations.
It is not LLM AI, it is just an ML model that rates the text content based on its themes. Just like your regular spam filter but a bit smarter 🐶🐾🫡
These abbreviations are jargon to me, must research this.
LLM is large language model, ai is artificial intelligence, ml is machine learning 🐶🐾🫡
Ah, I've progressed from a few capitalised enigmatic letters to many lowercase enigmatic letters, I'm so close! This is all so esoteric, how exciting!
This will happen sometimes, for a certainty. ML, like statistics, is all about the most efficient way to predict Y knowing X1, X2, X3 etc. If you're training a filter to block people whose musical tastes you don't like, and that is highly correlated with being black, the filter will totally use "black" as a shortcut to improve its accuracy. There are countermeasures, but the model can evolve new proxy variables for race much faster than its trainer can construct new reward schedules to penalize those. Cost of doing business, I'm afraid. Err on the side of tolerance, as much as your individual patience can stand.