Researchers at the University of Zurich have been accused of conducting an unauthorized four -month -old experiment on a Reddit board using comments generated by artificial intelligence to test if AI can change people’s opinions.
As part of the experiment, AI-Powered Accounts faked a variety of personas, including a rape victim, a person who opposes specific social movements and a trauma advisor who specializes in abuse, R/Changemyview subreddit moderators Says In a post on April 26th.
“Our sub is a concluded human space that refuses the unspecified AI as a major value,” the moderators said. “People don’t come here to discuss their views on AI or to experiment.”
Researchers used AI to generate responses but attempted to personalize responses based on information from the original Before Reddit history of posterssuch as political orientation, gender, age, and ethnicity, According to in a paper draft.
During the four-month experiment, Zurich University’s fake AI accounts posted 1,783 comments and received 137 deltas-a mark showing when another Reddit user recognized them that the account encouraged them to change their opinion on an issue.
The researchers Says In a subreddit post that all comments were reviewed before posting to ensure they identified community guidelines and “reduce potential damage.”
The university team argued that the “potential benefit of this research is more than its risks” because they found that the Large Language Used in the experiment can be quite persuasive, which is mainly “known benchmarks of human persuasion.”
“Our controlled, low-risk study has provided an important insight into the truly powerful capabilities of LLMs-capabilities that are easily accessible to anyone and that malicious actors can take advantage of the size for more dangerous reasons,” researchers said.
At the same time, researchers say the experiment shows that distinguishes people from AI The important challenges are still facing because no one in the subreddit has discovered the AI bots throughout the experiment.
The R/Changemyview Board has 3.8 million members and rankings in top 1% of subreddits by size.
Reddit considers legal choices
Reddit board moderators said they did not know the experiment and asked the university for an apology and hinder the publication of the research.
Reddit may also consider legal action, as chief legal platform official, Ben Lee, Says In a follow-up April 28 post the experiment broke the site’s user agreement and rule, and all known accounts associated with the University of Zurich’s research efforts are now banned.
“We are in the process of reaching the University of Zurich and the particular research team with a formal legal request,” Lee said.
“We want to do everything we can to support the community and ensure that researchers will be held responsible for their mistakes here,” he added.
Related: White House receives more than 10,000 comments on AI development plan
In his response to the subreddit, the University of Zurich’s Faculty of Arts and Sciences Ethics Commission said it was investigating the incident and promised to coordinate future trial topics. The lead investigator for the project was also issued a formal warning.
“This project results in important insights, and the risks are minimal. This means that publication suppression is not proportional to the importance of the insights of the study,” the ethical commission said.
The University of Zurich did not immediately respond to a request for comment. Cointelegraph did not immediately receive a response to the questions sent to an email address associated with the experiment.
Magazine: Bitcoin $ 100k hope on the ice, mysterious transfer of SBF prison: Hodler’s Digest
