The literacy paradox: How AI literacy amplifies biases in evaluating AI-generated news articles

This study explored how AI literacy can amplify biases when evaluating AI-generated news based on their content type (data-driven vs emotional).

Extracts:

· “Higher AI literacy can intensify opposing biases. When individuals better understand the use of AI in creating data-driven articles, they exhibit automation bias… Conversely, when AI generates opinion- or emotion-based articles, high literacy fosters algorithmic aversion”

· “AI literacy does not inherently enhance credibility; rather, it may paradoxically undermine it. This finding underscores a complex dynamic: while AI literacy aims to clarify AI operations and foster understanding, it may instead induce skepticism and diminish trust in AI-generated news content”

· An automation bias exists where “Individuals… reinforce the belief that AI excels at handling factual information with precision. This… strengthens the perception that AI is inherently objective, neutral, and reliable in data-driven contexts”

· An algorithmic aversion exists where “Individuals with high AI literacy are… aware of AI’s strengths but also its limitations, particularly its inability to comprehend human emotions, social nuances, and moral complexities”

· “AI literacy reinforces two distinct heuristics: automation bias and algorithmic aversion. Confirmation bias occurs when individuals selectively focus on and recall information that aligns with their preexisting beliefs”

· “efforts to improve AI literacy through education may end up further entrenching these biases, shaping audience perceptions of news credibility”

· In study 1 they found that high AI literacy increases the perceived credibility of AI driven news, decreasing credibility of interpretive AI news

· In a real-world chatbot experiment they found automation bias and algorithmic aversion mediate the effect of AI literacy on credibility judgments

· “High level of AI literacy increased both message and source credibility… when the AI author wrote a data-driven news. However… a high level of AI literacy diminished both message and source credibility… when the AI author wrote interpretive news”

· “AI news services should design their AI news chatbots to inform users about their algorithms during the conversation before presenting data-driven news articles… When presenting interpretive news articles… emphasize human-like qualities to enhance anthropomorphism”