'ai' idpol filter

Posting mode: Reply [Return]
Name
Options
Subject	Spoiler Image
Comment
Flag
File
Embed
Password	(For file deletion.)

File: 1672157642218.png (321.66 KB, 412x504, TN,XanaduStandModel,1983.png)

'ai' idpol filter Anonymous 2022-12-27 (Tue) 16:14:02 No.18068

i would like to discuss an idea thats been floating around in my head for a while, i haven't seen this brought up anywhere yet

trying to devise a way to actually organize effort around a common goal, primarily a primary defense against getting psy-op'd

the general idea is to make an ML algorithm, NN, what-have-you, that is designed to filter out common idpol/psyop from discussion and return concrete, actionable information. what's being suggested is essentially a great equalizer, it flattens the playing field so that the work necessary can be done without worrying who's glowing. typical glow tactics would be programmed in, and divisive messaging would simply come out as 'please remove me from the group cause i'm not in a position to actually help'.

for example, your org is going to do up a mutual-aid. someone still liberal would typically be crying and complaining about how they're oppressed black disabled lesbian muslim etc. etc. the algo would filter that to the recipient and turn it into something like, "I'm in a position where I need financial assistance before I can actually help with this thing". Then resources could be distributed to them accordingly, freeing them up to be actually useful.

maybe not the best example but interested in feedback

Anonymous 2022-12-27 (Tue) 16:28:16 No.18070

Learn what idpol means.

That said, we could at the very least train one on 4chan /pol/ and perhaps /mlpol/ so we can filter out actual idpol. It would at least get the Z posters to go away.

Anonymous 2022-12-28 (Wed) 09:11:02 No.18079

just have more not college ideologue sort mods who are from outside the West, that is the imperial core and NATO. intersectional radliberalism and spooked bourgeois academicism ironically works quite poorly on the real world's people of color as evidenced by the immense political incorrectness, on their spaces online, often demonstrated readily that yts almost couldn't fathom sometimes. no silly machine could copy the skills of tariq nasheed that easily, for example. or if still it is necessitated, train on tumblr's most popular topics and tags for discourse.

Anonymous 2022-12-28 (Wed) 10:21:58 No.18081

Anonymous 2022-12-28 (Wed) 19:44:30 No.18084

>>18068
Meds

Anonymous 2022-12-29 (Thu) 17:58:21 No.18092

the people you don't like are trying to use censorship to suppress "misinformation", right? in that case, your best bet would be to keep an eye on what kind of technology they develop to achieve their goal, copy it from them and apply it for your own goals. both sides want to control the narrative, so it makes sense that the means to achieve it would be the same.

Anonymous 2022-12-30 (Fri) 03:19:51 No.18096

>>18068
id-pol operates by self-referencing
you'd have to make a heuristic linguistic filter algorithm to detect self-referencing.

AI stuff uses a lot of computational power, i don't know if that's really viable.

The least effort variant is word-substitution for identities.

However be warned that if you make a censor system, all the identitarians are going to flock to it in order to let their id-pol pass through while they block the id-pol of the rest. You somehow have to find a mechanism that plays the different id-pol factions against each other so that they all block each others id-pol. Seems like a tricky thing to get right.

I don't know maybe it's better to train a gpt-bot on marx's collected works and use it to drown out idpol with in depth economic discussion.

Anonymous 2023-01-02 (Mon) 02:43:33 No.18113

Topic blocking is a cat and mouse game, but it could help.
Account blocking becomes a cat and mouse game but can be done.
If capital wants to obstruct something, they will make it so.

Unique IPs: 8