The Italian group “Informatics and Cultural Processes” has carried out four surveys of M people during four international AI conferences (1983–1985) This paper presents a synthesis of the results of these surveys along with a discussion from the point of view of a sociologist of knowledge. Developments in AI are important enough to raise questions about the future of human-machine relations, starting from the cultural premises of the Al community rather than considering progress in Alas a datum and a source of cultural consequences. It is found that AI people differ in a number of ways in their view of humans and their complexity. But this promising variety could be eclipsed by some achievements, such as expert systems, that have to some extent reduced the aims of AI to high-performance inferential problem-solving. On the other hand, regulatory feedbacks against a too formalized way of conceiving human thought could come from culture itself. The cultures of Europe and the United States seem to play different roles in the progress of AI— a difference that should perhaps be preserved as a resource until a satisfactory general theory is developed.