A study has exposed the startling vulnerability of AI-generated tweets, prompting none other than Elon Musk, who acquired Twitter in October 2022, to share his insights on this perplexing phenomenon.
What Happened: In a groundbreaking study published in the Science Advances journal, researchers have uncovered a concerning pattern: people are more susceptible to the influence of AI-generated tweets compared to those written by humans.
The findings have raised significant questions about our vulnerability to misinformation, sparking discussions on the role of AI in shaping public opinion. Musk weighed in on this disconcerting trend with a single word: “True.”
The study surveyed 697 individuals to determine their ability to distinguish tweets created by OpenAI‘s GPT-3 from those composed by humans. The results were startling, as most participants struggled to differentiate between AI-generated and human-crafted tweets. In fact, many actually found the AI-generated content more persuasive.
To gauge the impact of AI-generated misinformation, the researchers specifically focused on topics commonly associated with false narratives, such as vaccines, COVID-19, evolution, and 5G technology.
Participants were asked to identify the accuracy of each tweet, revealing a troubling outcome. Surprisingly, participants found it easier to discern the accuracy of AI-generated tweets containing genuine information. However, they were more likely to be deceived by AI-generated tweets containing deliberate falsehoods.
This led the researchers to label GPT-3 as a “double-edged sword” that can produce both reliable information and compelling disinformation.
Why It’s Important: This study corroborates existing concerns raised by experts regarding the potential dangers of widespread AI adoption, particularly in relation to misinformation.
Check out more of Benzinga’s Consumer Tech coverage by following this link.