AI Summaries Are Not Killing Reading, They Are Part of It
Linguist Naomi S. Baron argues that AI may weaken motivation to read for analysis, empathy, escape, and pleasure. However, the article counters that summaries have always been part of education: scholarly introductions, lectures, textbooks, and review essays all compress texts. AI summaries differ in speed and scale but not in basic function. Reading itself is a form of summarization, as described by Walter Kintsch and Teun van Dijk's model of text comprehension, where readers construct a mental 'macrostructure.' Memory research by Jacqueline Sachs and fuzzy-trace theory show that people retain gist over exact wording. The relevant comparison is between different forms of compression, not between reading and summarization. Summaries cannot replace reading for pleasure, language learning, close reading, or legal/medical work. A study of 195 participants found AI summaries helped lower-performing readers but reduced comprehension for stronger readers. An eight-week study showed students used AI summaries to triage texts. The educational task is to teach students when compression is appropriate and when the full text is necessary.
Key facts
- Naomi S. Baron worries AI makes books feel obsolete.
- Summaries have always been part of education: introductions, lectures, textbooks, review essays.
- Walter Kintsch and Teun van Dijk described text comprehension as building a mental macrostructure.
- Jacqueline Sachs found people quickly forget exact wording but retain meaning.
- Fuzzy-trace theory distinguishes memory for exact form and gist.
- A study of 195 participants found AI summaries helped lower-performing readers but hurt stronger readers.
- An eight-week study showed students used AI summaries to decide which parts to read deeply.
- The article argues the proper question is what kind of reading a task requires.
Entities
—