A new study from a professor at the Yale School of Management, Balázs Kovács, reveals that AI-generated restaurant reviews can pass the” Turing test” by deceiving both AI detection and human visitors, possibly undermining the trust of online reviews.
According to SuchScience, online reviews have become a key component of consumers ‘ decision-making process, with the majority relying on them to make wise choices. However, the development of powerful AI language designs is now threatening to undermine the reliability of these evaluations. Professor Kovács conducted two experiments with a different group of 301 participants, cut between the two studies, to check the capability of AI- generated reviews to mislead people readers and AI detectors.
Participants in Study 1 were shown both genuine Google reviews and artificially generated GPT-4 reviews created by OpenAI. Astonishingly, they correctly identified the cause only about 50 percent of the time – no better than random probability. Study 2, where GPT- 4 created totally imaginary reviews, yielded perhaps more impressive results: participants classified AI- generated reviews as individual- written 64 percent of the time.
Kovács likewise tested leading AI monitors to differentiate between AI-generated and human-written words. He fed 102 of the reviews to Copyleaks, a officially available AI- word recognition tool, which labeled all of the reviews as individual- generated, indicating its inability to identify the AI- generated material. When asked to determine the likelihood of each review being AI-generated, GPT- 4 was unable to effectively identify between human-written and its own AI-generated reviews.
The results have much- reaching implications for review programs, businesses, and consumers. Unscrupulous actors could use AI to create fake reviews, erode trust in online platforms, and have a disproportionately negative impact on small businesses that rely heavily on authentic reviews. The study serves as a wake-up call for policymakers to consider regulating transparency and for review platforms to reevaluate their authentication mechanisms.
Read more here at SuchScience.
For Breitbart News, Lucas Nolan reports on issues involving free speech and online censorship.
