This paper explores the following questions: What is reading all about, as our technical milieu becomes increasingly digital and our reading increasingly automated? What is entailed in closely reading a book, in studying and handling the book as an object? And what is the role of philosophy--and in reading philosophy--as we grapple with new technical modes of reading? Guided by philosopher Gilbert Simondon, this paper compares the language heuristics of large language models (LLM) with human reading practices, revealing parallel and diverging technical tactics, with the aim of increasing our understanding of how and why these algorithms are part of our technical reality. This comparison moves beyond concerns with automation and alienation, using Simondon's notions of technicity and transindividuality to philosophically analyze the nature of collaborative reading in a distraction economy, and the extent to which transformer neural network models achieve an implicit embodied or grounded sense of language-use.