The dizzying possibilities of our age expose how far our technology has outrun our ethics—how our ability to do things has overwhelmed our ability to think about whether we should do those things. As with various biotechnologies, artificial intelligence has raised questions our society is not prepared to answer. Sadly, for the most part, neither is the Church.  

In a scene from the PBS animated series Arthur, Buster the rabbit asks in shock, “You really think someone would do that? Just go on the internet and tell lies?” Obviously, Buster was a little naive. In the 20 years since, the internet has proven to be a bottomless source of confusion, propaganda, and misinformation.  

Until recently, online lies have come from actual people. This is no longer the case. A team of researchers at the University of Zurich recently performed a highly questionable experiment with AI bots impersonating people. Apparently, they were very effective at changing the beliefs of people on Reddit.   

In their paper, entitled “Can AI Change Your View? Evidence from a Large-Scale Online Field Experiment,” these researchers detailed how they tasked AI bots to study user profiles without their knowledge in order to find vulnerabilities. The bots then posed as real people in chat forums and tried to persuade human users of left-leaning beliefs. It worked frighteningly well.  

The researchers estimated that the AI bots, some of which pretended to be rape victims, LGBTQ, or government employees, achieved a persuasion rate six times that of human users. On topics ranging from abortion to whether Christianity is good for the world, the bots deployed arguments fine-tuned to exploit vulnerabilities, often using lies, misinformation, or highly debatable claims. And again, it worked. After conversing with the undercover bots, people changed their minds. 

As tech entrepreneur Mario Nawfal put it, this experiment “should chill anyone who values authentic human discourse.” These “digital ghosts” crossed “critical ethical lines,” fabricating over 1,500 comments, “each precisely calibrated to exploit cognitive vulnerabilities of their human targets.” Also, this was done without the knowledge or consent of those targeted. 

Read More