Abstract

The successes of recent algorithms based on machine learning and Big Data are often presented as the realization of the project of Artificial Intelligence, reproducing with machines the processes of human intelligence. Instead, the article argues that systems such as ChatGPT work so well not because they have learned to become intelligent, but because they have learned to produce contributions that enable their users (we intelligent humans) to produce our relevant and interesting information, i.e., they have learned to participate in communication. The interaction with machines will have profound effects on the development of intelligence, but in order to grasp them we should rather move away from the comparison and competition between the cognitive abilities of humans and machines