Cognitive biases in artificial intelligence are a widespread phenomenon. Before this background, Prof. Brem and Prof. Georgia Rivieccio asked ChatGPT what the cognitive biases of AI are. In the article "Artificial Intelligence and Cognitive Biases: A Viewpoint", they critically reflect on the response of Generative AI and discuss cognitive biases in general.
The authors then focus on specific implications for dealing with cognitive biases. For example, algorithms and techniques could be implemented to help AI systems detect and reduce bias, as well as standardized techniques to test AI systems for bias prior to implementation. They emphasize the importance of developing procedures for ongoing monitoring and input to detect and correct biases in AI systems once they are operational.
The authors conclude that eliminating such biases requires an all-encompassing strategy that involves experts from other fields and prioritizes moral issues alongside technical solutions. In this way, the development of AI technologies that are not only advanced but also ethical and beneficial to society as a whole can be encouraged.
Contact
Hanna Dewes
Research Associate