Wikipedia Becomes A Battleground For Detail-Obsessed Bots

September 23, 2016 - 2 minutes read

microsoft-tay

It turns out bots aren’t as peace-loving as Microsoft’s Tay might have lead us to believe. (I jest.)

According to a research study published by Cornell University, Wikipedia has become a quiet, passive-aggressive battleground for communities of bots deployed to conduct routine maintenance and undo the work of socially-motivated trolls. Rather than concentrate on human errors, bots have spent long periods of time — sometimes years — reverting and re-reverting minor details in articles throughout Wikipedia.

The thesis of the paper is that relatively simple bots are displaying surprisingly complex behaviors — which could have compelling implications as AI and deep learning become integrated into everything from mobile apps to IoT devices.

Mobile app developers are taking note, particularly as the chat bot trend is taking off in walled-garden mobile app ecosystems like Messenger and Kik. Sure, the differences the bots squabble over are so inconsequential as to go unnoticed by human users. Regardless, the fact that such simple programs are displaying unexpected behavior warrants a closer look.

Perhaps most interesting about the study is the statistics on bot activity in different regions and languages. English-speaking bots, for example, are far more “argumentative” than other languages. Portuguese bots are shown to have far more disagreements than german ones, as measured by the number of “reverts” performed.

The researchers take a dark tone in spite of the amusing nature of the subject, warning NYC app developers that “t is crucial to understand what could affect bot-bot interactions in order to design cooperative bots that can manage disagreement, avoid unproductive conflict, and fulfill their tasks in ways that are socially and ethically acceptable.”

As AI bots become deeper embedded in borderline-universal products like Facebook and Google, it’ll be fascinating to see how similar complex behaviors affect the mobile app development community.

Tags: , , , , , , , , , , , , , , , , , , , , , , , , , , , ,