In Reality How AI Changes Voters' Minds with Oxford University's Kobi Hackenburg
Welcome to In Reality the podcast about truth disinformation and the media with Eric Schurenberg, long time journalist and media executive, now the founder of the Alliance for Trust in Media.
Up to this point, Eric's assumption about artificial intelligence—and he's not alone in this—has been that it will only make the information environment worse. More chaotic, more polluted by mass produced slop that only reinforces our partisan belief bubbles. Add to that the scary possibility is that AI would become a superhuman tool of political manipulation, able to strike up one-on-one conversations with voters about whom the AI knows everything and win them over with subtle manipulation.
It so happens that three scientific papers have come out just last week that challenge this alarming scenario. All concluded that AI models in conversation are indeed much more persuasive than conventional methods used to sway voters—like, say, political ads. But they are most effective not because they are especially manipulative, but because they are better at marshalling facts to support their case.
Today’s guest is Kobi Hackenburg, an Oxford University PhD candidate who authored one of the papers and who has long studied the question of AI’s power to change minds. Among other things, Kobi and Eric will dig deeper into why factuality works so well in these experiments and other, more insidious techniques, like microtargeting, don’t. We’ll speculate about how AI might be used by political campaigns in the future, And whether we should be cheered ar that, at least in this experiment, people seemed surprisingly open to changing their minds when confronted with contrary facts.
Website - free episode transcripts
www.in-reality.fm
Alliance for Trust in Media
alliancefortrust.com