Feb. 23 (UPI) -- An analysis of bot behavior over the course of a decades shows even "benevolent" bots bicker. In fact, researchers found evidence of bot-versus-bot fights lasting several years.
The bots in question were employed by Wikipedia to perform a variety of editing and maintenance tasks. Editing bots repair vandalized text, enforce content and language bans, check and fix spelling, insert links and more. Other non-editing bots scan text for copyright infringements and other violations.
In a survey of bot activity between 2001 and 2010, researchers found evidence of bot interactions. Though the bots were not specifically designed to interact with one another, they did.
"We find that, although Wikipedia bots are intended to support the encyclopedia, they often undo each other's edits and these sterile 'fights' may sometimes continue for years," researchers reported in PLOS ONE.
Researchers looked at bot-bot interactions on different language versions of Wikipedia. On English Wikipedia, bots undid another bot's edits an average of 105 times over 10 years. Portuguese Wiki bots were the most meddlesome, averageing 185 bot-bot reverts per bot.
Researchers say their findings reveal bot-bot interactions to be more unpredictable than previously thought. The research has implications for artificial intelligence used in other capacities, like cybersecurity or autonomous vehicles.
"We tend to forget that coordination even among collaborative agents is often achieved only through frameworks of rules that facilitate the wanted outcomes," Luciano Floridi, a professor at the Oxford Internet Institute, said in a news release. "This infrastructural ethics or infra-ethics needs to be designed as much and as carefully as the agents that inhabit it."
Even if the algorithms that govern bots are relatively simple, bot-bot interactions can yield complex ecosystems.
"We find that bots behave differently in different cultural environments and their conflicts are also very different to the ones between human editors," Oxford researcher Milena Tsvetkova said. "This has implications not only for how we design artificial agents but also for how we study them. We need more research into the sociology of bots."