KEB: boti, kteri delaji zamerny preklepy a podobny non-grammar-compilant veci, jsou afaik celkem bezna vec
imho a zrovna na nejaky carky v souveci nepotrebujes zadny komplexni AI/ML/LLM ... na to ti staci randomizovany nahrazovani textu - maly velky pismenka na zacatku vet, pravdepodobnostni prehazovani interpunkce a mezery, etc..
kdyz jsme u tech botu a gramatickych chyb
vypadl na me zajimavy paper
tady sice preklepy nejsou generovany botama, ale pomahaji s jejich amplifikaci
Social Media Bots: Implications for Special Operations Forceshttps://apps.dtic.mil/sti/trecms/pdf/AD1112595.pdf(DTIC = Defence Technical Information Center)
Fracturing: breaking a large conversation into smaller conversations
Fracturing occurs exclusively around social media hashtag campaigns. In these cases, the
erceding entity deploys a purposefully misspelled hashtag to break a larger conversation
into a series of smaller conversations. These “misspelled hashtags are decoys, aimed at
diffusing the reach of the original by breaking the conversation into smaller groups.
.....
In supporting these misspelled hashtags, the Russian botnet aspired to “train Twitter’s search
algorithms to see the misspelled versions as trending topics.” And then, because they were
trending, Twitter began to suggest them to users who were searching for the correctly spelled
hashtag.
....
Although it is true that the effective use of social media bots may require cultural expertise in
some instances (unlike for the use of memes), it is not universally true. In some cases, social
media bots and botnets fall outside this rule and provide a means of influencing conversations
with no cultural expertise. For example, a trending hashtag or social media account could be
flooded with content that is completely meaningless (e.g., photos of giraffes) or fractured by
simply changing one letter in the spelling of the hashtag. In each instance, it is not necessary to
have a sophisticated understanding of the target audience to achieve the desired effect.