If we are concerned about the consequences of using algorithms on content such as text, sound, and video, it is because content has become calculable. By considering only their syntactic materiality, these forms of content are translated into formats, that is, into formal tokens to be manipulated solely based on their syntactic type, independently of their meaning. In this sense, AI is the project of translating content into algorithms that generate meaningful and relevant transformations, despite their restriction to formal structures. After outlining the historical and conceptual roots of formalisation as a means to render content computable, this chapter reviews the main classes of tools used in AI. It concludes with a discussion of the status of the results produced by these tools, particularly the content created by generative AI. This content appears meaningful, and thus seems to address us and represent the world, which in turn raises fundamental questions about enunciation (who is speaking) and modelling (what world is being represented). Upon closer examination, generative AI appears to offer a calculable meta-archive of everything that has been said, so that it can be said again, differently.