Journée d'Etude : Genre et Intelligence Artificielle
Institut Nicod - Défis GENRE - CNRS
Vendredi 10 juin 2016
Thirty years of gender studies have resulted in a new awareness of a social, biological and cultural dimension of our life that must be rethought and reconsidered. Yet, although the impact of these studies on social life and policy-making is obvious, there are many questions that are left to explore:
The problem with the study of gendered differences is that our folk psychological concept of gender and its social implications deeply influence the process of epistemological and semantic construction of an abstract object of inquiry. Indeed, as Cecilia Ridgeway has noticed “it is striking that people find it almost impossible to relate to a person that they cannot classify as male or female”. We don’t even know how to address a person we can’t place as male or female. That is why so many social stereotypes on gender are uncritically imported into scientific psychological research although “objectivized” through a process of decontextualization and abstraction.
Addressing the question from the angle of Artificial Intelligence is a fruitful way to tackling these issues: What does a “gendered intelligence” mean? Is it possible to conceive a gendered human-like machine without incorporating the existing stereotypes we have about gender? Philosophers, psychologists, AI experts and artists will gather for a day to think about these issues.
Informations pratiques :
La journée sera en anglais. Elle est ouverte à tous, dans la limite des places disponibles. Pour vous inscrire, envoyez un email à Gloria Origgi
11:30 - 13:00 (co-organisé avec le Colloquium de l’IJN) - Salle de réunion de l’Institut Nicod, Pavillon Jardin - RDC, 29 rue d’Ulm 75005 Paris.
Friederike Eyssel (Bielefeld University)
What's social about social robots?: The case of gender
This talk proposes a social psychological perspective on what makes robots appear "human". A series of experimental studies will be presented to document the key role of core social categories in impression formation processes about nonhuman entities. The primary focus will be on the aspect of gender (i.e., user-gender, gender-typicality of a task or context, and robot gender) to provide insights into social-cognitive processes and judgments about gendered technical systems. Implications for basic and applied research in psychology and robotics will be discussed.
13:00 - 14: 00 Lunch, Buffet devant la salle Prestige 1, Ecole normale supérieure, RDC, 29 rue d’Ulm, 75005
14:00 - 15:00 - Salle Prestige 1, RDC, Ecole normale supérieure, 29, rue d’Ulm 75005 Paris.
Rawan Charafeddine (Institut des Sciences Cognitives Marc Jeannerod, Lyon)
Dominance suits the boys better: Young children’s expectations about gender and dominance
Dominance relationships are at the core of gender system and even the most advanced countries are yet plagued with gender inequality. To assess preschooler’s inferences about dominance and gender, we conducted three intercultural studies and obtained the following results:
1) Preschoolers infer masculine gender from dominant status, regardless of their age, sex and culture.
2) Girls are less likely than boys to attribute the dominant status to themselves in imaginary mixed-gender situations.
3) Reasoning about concrete situations seems to drive some of the girls, rather than boys, to revise their stereotypical knowledge in this domain.
4) Boys are more likely than girls to attribute higher status to masculine individuals.
15:00 - 15:30
Fabienne Audeoud (artiste) : Présentation d’une vidéo originale créée pour la journée et intitulée : A conversation about Gender and AI
15:30 - 16:30
Gloria Origgi (Institut Nicod)
A Turing Test for Gender
In 1950, Turing proposed an imitation game in which a man and a computer compete in pretending to be a woman. The Imitation Game asks a computer to not only imitate a thinking human, but a specific gender of thinking human. The initial setting of the Imitation Game was the following: There are two individuals pretending to be a woman, one is a man and the other is a woman. And there is an interrogator, whose gender is unimportant, who tries to guess who is the woman. Then Turing asks: “What if we replace the man by a machine in this game”? The gendered dimension of the Imitation Game has been evoked in many discussions about of gender theory (cf. J. Genova 1994). Yet, surprisingly, the gendered dimension of the test has not been exploited in order to write “scripts” that could fool an interrogator about the gender of the participants. How should such a script conceived? If we stick to Turing’s idea of testing only “cognitive” features of the subjects and let out the physical ones, what could be an adequate script for the Gendered Imitation Game? Examples of scripts will be given and discussed.