Impact of AI on Human Perception
-Generative AI models, including notable examples like ChatGPT, DALL-E, and Midjourney, have raised concerns about their potential to distort human beliefs by transmitting inaccurate information and perpetuating stereotypical biases, according to a perspective shared by researchers Celeste Kidd and Abeba Birhane. Exploring the intersection of human psychology and generative AI, they examine the underlying factors contributing to the influence of these models on shaping human beliefs.
Overestimation of AI Capabilities
-Kidd and Birhane argue that society’s perception of the capabilities of generative AI models has been inflated, fostering a widespread belief that these models exceed human capacities. This tendency arises from the natural inclination of individuals to embrace information from knowledgeable and confident sources, such as generative AI, more readily and with increased certainty.
AI’s Role in Disseminating False and Biased
-Information Generative AI models possess the potential to generate and spread false or biased information extensively, with the repetitive nature of this dissemination playing a pivotal role in solidifying such information within people’s belief systems. Individuals are particularly vulnerable to influence when seeking information and are more likely to maintain a firm grip on the received information.
Implications for Information Retrieval and Provision
-The current design of generative AI primarily caters to information retrieval and provision, posing significant challenges in countering false or biased information once individuals have been exposed to it through these AI systems, as highlighted by Kidd and Birhane.
The Need for Interdisciplinary Research
-In conclusion, the researchers emphasize the urgent need for interdisciplinary studies to comprehensively evaluate the impact of generative AI models on human beliefs and biases, both before and after exposure to such models. This timely opportunity becomes increasingly crucial as these systems are progressively integrated into various everyday technologies.
Reference: “How AI can distort human beliefs: Models can convey biases and false information to users” by Celeste Kidd and Abeba Birhane, Science.