Social media accounts controlled by computer bots (bot accounts) have become increasingly common on many social media platforms and messaging apps. The number continues to grow benefiting from generative AI technologies such as the Large Language Model (LLM) since 2022.
As a result, it’s getting easier and cheaper to create AI content using thousands of grammatically correct and convincing bots following different situations. “It’s now much easier to customise these big language models for a specific audience with a specific message,” said Paul Brenner of the Centre for Computational Research at the University of Notre Dame in Indiana.
See also:
3 Ways to make AI Hug videos that go Viral on TikTok and Instagram
Brenner and his colleagues found that respondents ‘ ability to distinguish AI-powered bots from humans was only an average of 42 percent. The ability was not greater even though the respondents had been told that they might interact with bots.
Tempo tried to be one respondent in the research as it is on the site nd.qualtrics.com. there, Brenner provided a situation in the form of news about leaks about AI regulation in the European Union that would free open source models from strict rules.
Then there were five netizens ‘ responses to the news. Respondents were asked to identify which responses came from bots, assess how comfortable or uncomfortable it was when reading their interactions, and write an explanation of how they could tell the difference. Of those five responses, Tempo guessed two of the three correct ones from the AI bot.
- Emojis and hashtags: their excessive use could be a sign
Advertising
Scroll To Continue
- Unusual phrases, diction, or analogies: the use of unusual words may indicate an AI bot
- Repetition and structure: bots may use repetitive words that follow similar or rigid forms and they may overuse some slang terms.
- Asking: this can reveal the bot’s lack of knowledge about a topic–especially about local locations and situations.
- Assuming the worst: if a social media account is not a personal contact and their identity has never been clearly validated and verified, it could be an AI bot account.