A study demonstrates that over 2,000 conspiracy believers interacted with the AI model DebunkBot, resulting in a 20% decrease in belief confidence. Participants, who initially firmly held conspiratorial views, experienced notable shifts in belief after roughly eight minutes of AI-led dialogue. This interaction showed lasting effects, with follow-ups revealing sustained reductions in conspiratorial belief two months later.
Significantly, the AI was able to provide accurate, fact-based rebuttals, effectively addressing common misconceptions surrounding various conspiracy theories. The ease of access to factual information has been hampered by the overwhelming volume of content available online, which makes discerning truth difficult. This research highlights the potential for generative AI not only as a tool for misinformation but also as a means of restoring rational discourse by delivering facts in an engaging format.
👉 Pročitaj original: MIT Technology Review
