Writing in Parallel

There’s a simple idea at the center of Writing in Parallel that might take you by surprise: most of AI’s problems are writing problems. Bill Resimont’s concise argument is that language models fail not because they’re bad at processing text, but because the text they’re trained on is bad — scraped indiscriminately from the internet, unverifiable, unstructured. His proposed fix is elegant: pay many writers to respond to the same prompt independently, preserve every voice exactly as it is, and let AI make sense of the collection.

He calls this parallel writing, and it’s older than you’d expect. The book traces the practice back to ancient Greek classrooms where students carved independent answers into wax tablets simultaneously, each working alone, each contributing to a shared picture of collective thought. What’s new is the technology to read all those tablets at once, at scale, and find the patterns without erasing the distinctions.

Resimont writes with a teacher’s instinct for structure. Each of the 12 chapters lays its foundation carefully before building up, and every chapter closes with 10 discussion prompts — this book clearly pulls double duty as course material. The prose is clear but methodical; readers looking for narrative momentum may feel the pace, but readers who want the argument built rigorously will appreciate the care.

The book is at its best when it engages directly with where AI has already failed. Microsoft’s Tay chatbot, Amazon’s gender-biased hiring algorithm, IBM Watson’s flawed oncology recommendations — Resimont walks through each case not as a cautionary tale but as evidence: the problem was the data, and the fix requires intentional, structured, human-sourced text. His framing is precise: “patterns become visible, blind spots emerge, and the reasoning behind each idea is preserved rather than flattened.” Parallel writing, he argues, can produce exactly that.

Writers, educators, and technologists who believe the next phase of AI requires more human intentionality, not less, will find Writing in Parallel a clear-headed and genuinely useful read.