Boredom is what separates us from machines: AI study of Twitter bots reveal

According to a study of human and fake Twitter accounts, our very short attention spans is what separates us from robots. Researchers from the University of Southern California have discovered this after analysing activity patterns in two Twitter data sets.

The first set was a set of posts about the 2017 French presidential election. The second was a collection of tweets from bots and human (real) accounts that has been hand-labeled. Researchers chose these two data sets because they complemented each other. The hand-labeled accounts differentiated between humans and bots reliably while those from the French election offered contextual insights.

Researchers analysed behaviours of these accounts by studying lengths of tweets, number of retweets, replies and mentions these tweets attracted.

Subsequent results revealed that while human interactions increased over time, humans also tweeted less. The bots, on the other hand, maintained the same level of activity and produced content at specific times - like 30-minute intervals.

According to researchers, this pattern exists because humans get too tired of creating original content and get distracted by other content. And they used this to train a bot-detection algorithm called Botometer that distinguishes between human accounts and fake accounts. When the AI ignored the timings of the posts, it detected bots better.

Insights like these can help AI detect fake accounts seeking to influence elections, at least till bots learn to mimic the human tendency of getting bored. However, the study used data from 2017, who knows what the bots might have learned by now.