An analysis of millions of tweets reveals the changing meaning of the word “bot”
Svet foto/Shutterstock
Calling someone a bot on social media once meant you suspected they were actually a piece of software, but now the use of the word is shifting to become an insult to someone you know is human, say researchers.
Many efforts to detect social media bots use algorithms to try to identify patterns of behaviour that are more common in automated accounts controlled by computers, the traditional meaning of a bot, but their accuracy is questionable.
“Most recent research really focuses on the detection of social bots, which is problematic in itself because we have this ground truth problem,” says Dennis Assenmacher at Leibniz Institute for Social Sciences in Cologne, Germany, meaning we don’t know if detections are accurate.
To investigate, Assenmacher and his colleagues looked at how users perceive what is a bot or not. They did so by looking at how the word “bot” was used on Twitter between 2007 and December 2022 (the social network changed its name to X in 2023, following its purchase by Elon Musk), analysing the words that appeared next to it in more than 22 million English-language tweets.
The team found that before 2017, the word was usually deployed alongside allegations of automated behaviour of the type that would traditionally fit the definition of a bot, such as “software”, “script” or “machine”. After that date, the use shifted.
“Now, the accusations have become more like an insult, dehumanising people, insulting them, and using this as a technique to deny their intelligence and deny their right to participate in a conversation,” says Assenmacher.
While the cause of the shift is unclear, Assenmacher says it could be political in nature. The team looked at high-profile accounts followed by each Twitter user, such as those of politicians and journalists, to classify users as left or right-leaning. It turned out that left-leaning users were more likely to accuse others of being bots, and those who were accused were more likely to be right-leaning.
“A potential explanation might be that media frequently reported about right-wing bot networks influencing major events like the [2016] US election,” says Assenmacher. “However, this is just speculation and would need confirmation.”
Topics: