A Writer Lets AI Absorb Typos but Keeps Revision

A writer who describes themselves as having dyslexic tendencies says they now let LLMs absorb their spelling errors, because the models read text as tokens and infer meaning from context, so typos such as "writel etters" no longer block understanding. The writer says the relief came at a cost: they stopped rereading and fixing their own mistakes. To keep from losing the ability to communicate without a model, the writer is reclaiming the work of revising and structuring arguments, while still leaving spelling errors to AI. The account is the writer's own description of their practice, not a measured result.

สร้างจากรายงาน · อัปเดต 6 ชม. ที่แล้ว

ยังมีการเฝ้าดูต่อเนื่องไม่พอจะวาดแนวโน้ม