Artificial intelligence: Relevance for developmental psychopathology

Over the past 10 years, psychiatry, psychology and artificial intelligence (AI) have opened a fruitful dialogue. In the field of social interactions and social signal processing, many multidisciplinary works have contributed to a better understanding of the complexity of certain psychological phenomena and to the progress of human-machine interfaces. They allow the computational consideration of the dyadic dimension, the possibility to take into account the multimodal nature of turn taking and the development of specific tools to accurately and automatically extract the social signals of partners. We propose here a brief survey of our work illustrating the interest of not only associating AI and psychopathology to explore certain specific contexts (effect of oxytocin, early maternal neglect, developmental robotics) and certain psychopathologies (autism, borderline states) but also to design therapeutic applications such as serious games or intelligent companion robots.