1. médialab Sciences Po
  2. News
  3. AI critiques in the press

AI critiques in the press

On the occasion of the 4S conference (October 6th to 10th in Toronto), the medialab publishes the first results of a research aimed at exploring critical discourses on AI in the media sphere. An interactive website offers an exploration of the data from this study.

Post

For years, artificial intelligence has been the subject of constant debate in the public space. For some, it is a vector of progress and transformation, but for many, its development raises concerns and criticism. Are these "intelligences" autonomous and what could be the consequences of their actions in the field of work, war, information or sexuality? Or are these entities rather a new equipment of our socio-technical environment intended to help the decisions of the humans by directing the conducts, by selecting information or by proposing probabilities on the occurrence of future events?

Using natural language processing applied to a press corpus,this research explores critical discourses on AI in the English-language media sphere. The semantic analysis of the corpus reveals two dominant regimes of criticism, which focus:

  • on the one hand on the injustices produced by the algorithms that shape our digital environments,
  • and on the other hand on the fear of robots and autonomous AI associated with a prophetic discourse.

The fear of robots and the criticism of algorithms constitute two types of discourse with very different characteristics. The first projects into the future representations that oppose humans and machines in general, while the second, anchored in the present, highlights a set of situations of discrimination, bias and neglect of which specific populations are victims. The opposition between these two registers allows us to observe the way in which an old cleavage in the history of the articulations between computer science and society is perpetuated and replayed. Should we think of artificial machines as "intelligent" entities in themselves, or are they tools for making humans and our societies more "intelligent"?

Legend : Semantique network of IA critical articles
Legend : Semantique network of IA critical articles

The website "Critique of AI in the press" proposes to explore the data by combining descriptive analysis and data visualizations (graphs, semantic networks, etc.). It presents the different stages of data processing and the methods used to analyze the corpus of press articles. 

* This research was supported by the Chair «Good In Tech : repenser l’innovation et les technologies comme moteurs d’un monde meilleur pour et par l’humain ».