An evaluation of hundreds of thousands of tweets reveals the altering that means of the phrase “bot”Svet foto/Shutterstock
Calling somebody a bot on social media as soon as meant you suspected they have been really a bit of software program, however now the usage of the phrase is shifting to change into an insult to somebody you already know is human, say researchers.
Many efforts to detect social media bots use algorithms to attempt to establish patterns of behaviour which can be extra widespread in automated accounts managed by computer systems, the standard that means of a bot, however their accuracy is questionable.
“Most up-to-date analysis actually focuses on the detection of social bots, which is problematic in itself as a result of we’ve got this floor fact drawback,” says Dennis Assenmacher at Leibniz Institute for Social Sciences in Cologne, Germany, that means we don’t know if detections are correct.
To analyze, Assenmacher and his colleagues checked out how customers understand what’s a bot or not. They did so by how the phrase “bot” was used on Twitter between 2007 and December 2022 (the social community modified its identify to X in 2023, following its buy by Elon Musk), analysing the phrases that appeared subsequent to it in additional than 22 million English-language tweets.
The workforce discovered that earlier than 2017, the phrase was normally deployed alongside allegations of automated behaviour of the sort that may historically match the definition of a bot, equivalent to “software program”, “script” or “machine”. After that date, the use shifted.
“Now, the accusations have change into extra like an insult, dehumanising individuals, insulting them, and utilizing this as a method to disclaim their intelligence and deny their proper to take part in a dialog,” says Assenmacher.
Whereas the reason for the shift is unclear, Assenmacher says it might be political in nature. The workforce checked out high-profile accounts adopted by every Twitter consumer, equivalent to these of politicians and journalists, to categorise customers as left or right-leaning. It turned out that left-leaning customers have been extra more likely to accuse others of being bots, and people who have been accused have been extra more likely to be right-leaning.
“A possible rationalization is perhaps that media steadily reported about right-wing bot networks influencing main occasions just like the [2016] US election,” says Assenmacher. “Nevertheless, that is simply hypothesis and would wish affirmation.”
Matters:
