New research conducted by psychologists LucÃa Vicente and Helena Matute from Deusto University in Bilbao, Spain, has uncovered a startling revelation: humans can inherit biases from artificial intelligence (AI). Biases, which are systematic errors in AI outputs, have the potential to influence human decision-making and lead to potentially harmful consequences.
The recent advancements made in AI technology have raised expectations for its reliability and accuracy. Various professional fields have started incorporating AI-based tools to assist specialists in decision-making processes and minimize errors. However, this reliance on AI comes with inherent risks, as biases in AI results have been observed. It is important to note that AI models are trained using data reflecting past human decisions. If this data contains patterns of systematic errors, the AI algorithm will learn and reproduce these errors, thereby perpetuating biases.
The groundbreaking finding of Vicente and Matute’s research is that humans may not only inherit biases from AI but also become trapped in a dangerous loop where biases are perpetuated. The study, published in Scientific Reports, conducted three experiments involving volunteers performing a medical diagnosis task. One group of participants was aided by a biased AI system, which exhibited a systematic error, while the control group performed the task without any AI assistance.
The results were astonishing. Participants who were assisted by the biased AI system made the same type of errors as the AI itself, whereas the control group did not exhibit these mistakes. Consequently, AI recommendations significantly influenced the decisions of the participants. However, the most significant finding was that even after the participants switched to performing the diagnosis task without AI support, those who had previously relied on the biased AI continued to replicate its systematic error. This suggests an inherited bias effect, which was not observed in the control group.
These findings raise concerns about the long-lasting negative impact of biased information from AI on human decision-making processes. They underline the urgent need for further psychological and multidisciplinary research on the interaction between AI and humans. Additionally, evidence-based regulation is essential to ensure fair and ethical AI, taking into account not only the technical aspects but also the psychological aspects of AI-human collaboration.
The study conducted by Vicente and Matute emphasizes the importance of understanding the potential consequences of biases in AI and their influence on human decision-making. It sheds light on the need to address these biases to develop more reliable and unbiased AI systems. By doing so, we can minimize the risks associated with inherited biases and ensure that AI remains a tool that supports and enhances human decision-making rather than hindering it.
In conclusion, the research conducted by Vicente and Matute has highlighted the significant impact of biases in AI on human decision-making. It has provided compelling evidence of the potential for humans to inherit these biases from AI systems, emphasizing the need for further research and evidence-based regulation. As artificial intelligence continues to advance, it is crucial to address the issue of biases to ensure fair and unbiased decision-making processes in various professional fields.
Reference:
Vicente, L, & Matute, H. Humans inherit artificial intelligence biases. Scientific Reports (2023) https://doi.org/10.1038/s41598-023-42384-8
