Using various methods of natural language processing on a large corpus of press articles covering a period of 5 years, we show two dominant regimes of criticism of artificial intelligence that coexist within the media sphere, involving different technological and human entities, but also distinct temporality and issues. We observe a shift between articles featuring algorithmic calculation techniques incorporated into the user's environment to guide, orient or calculate his behaviors, towards articles characterized by a personification of AI in an embodied and autonomous entity. This topology can be interpreted as a process of progressive independence of AI that constitutes a constant polarity in the history of the relationship between computer technology and society. On the one hand, robots and AI, which refer to autonomous and embodied technical entities, are associated with a prophetic discourse alerting on the capacity to control these agents that simulate our physical and cognitive capacities and threaten our physical security or our economic model. This regime of critical enunciation is organized around the feeling of fear regarding the autonomy of robots, which thrives on science fiction, religious concerns and contains a prophetic dimension on the future of humanity. On the other hand, the technical entities are represented by algorithms that shape our daily computing environments. They are associated with a short term discourse, focusing on populations characterised by their age, ethnicity, political or sexual orientation, and mobilizing a social justice criticism that denounce bias, discrimination, surveillance, censorship and amplification in the diffusion of inappropriate content.