The aim of this article is to explore what can be expected from generative AI in terms of creativity in the case of text, and bodily gestures in the case of image production - sound and music are only addressed laterally. The article argues that the power of generative AI is primarily based on linguistic commands embodied in textual prompts, the use of which runs directly counter to the metaphorical capacity of language, which cannot be measured statistically, as any literary text demonstrates. It is therefore the implicit transformation of what we can expect from language that lies at the heart of the question of the power of the new generative AI.