Behavioral Data Science has revolutionised the development and application of Artificial Intelligence (AI) through the development of new human-algorithm systems and environments, where humans better understand the logic behind algorithmic decision making and algorithms are better equipped to understand and serve humans.
The interdisciplinary field of Behavioral Data Science combines techniques from psychology, economics, sociology, and business, with computational approaches from computer science, statistics, data-centric engineering, information systems research, and mathematics to model, understand, and predict behaviour. Behavioral Data Science addresses human behaviour, algorithmic behaviour, and systems behaviour. The human behaviour strand provides methodological tools that use research in psychology, decision theory, and behavioural science to explain human behaviour using large datasets. The algorithmic behaviour strand combines algorithmic methods from statistics, computer science, and mathematics to predict and explain the behaviour of humans and systems using algorithms as well as to better understand the machine and algorithmic behaviour. The systems behaviour strand incorporates methods that allow the modelling of complex systems, networks, markets, cultural differences, and culture in various contexts.
The integration of behavioural data science methods into AI has revolutionised the field by enabling the development of more accurate, explainable, and ethical AI systems that are better equipped to understand and predict human behaviour. By combining insights from behavioural science with advanced computational techniques, such as machine learning and deep learning, researchers and practitioners are now able to develop more functional and effective AI systems that can address a wide range of societal challenges while also promoting individual and collective well-being.
One of the ways that Behavioral Data Science has revolutionized AI is by addressing the need to study behaviour "in the wild," outside of the controlled laboratory setting to develop more usable complex human-algorithm systems. Through this work, Behavioral Data Science has contributed to the development of more functional systems and algorithms that better understand humans. By combining behavioural science models with AI algorithms, it is possible to improve and simplify predictions of human behaviour in a wide variety of contexts.
For example, the anthropomorphic learning approach allows the development of more accurate suggestion systems, traceable and explainable models, and better predictions of human, algorithmic, and systems behaviour. It requires smaller training sets and often outperforms existing deep learning algorithms. One of the reasons why AI algorithms tend to ignore existing insights from decision theory and behavioural science is because these models tend to operate as black boxes, providing predictions without a clear understanding of what drives people's preferences. In contrast, hybrid models, such as anthropomorphic learning, provide more insights into people's behaviour by combining deep learning models with behavioural science models. This combination allows for more accurate and interpretable predictions and can provide a better understanding of what factors drive human decisions. Refer to the Impact section of this website for various applications and use cases of Behavioural Data Science.
Selected References
Comments