Abstract
The development of flexible automation and computerization in industry and the services brings to light the question of the nature of human work and of skills in automatic systems. Many studies have had a tendency to consider work either as a residual form of activity or, on the contrary, as a factor of resistance to change. The analysis of what work consists of in automatic systems actually reveals a great variety of human monitoring processes because these systems are very often only partly automatic or computerized. This results in the constant use of human intelligence in order to identify and find solutions for incidents and mishaps. These tasks imply a cognitive approach based on the learning of computer languages, codes, incidents and malfunctions which enables workers to control automatic systems and technical development.