I keep running into this weird thing where AI can make a draft, answer, or workflow step look way more complete than it really is. The output isn’t always wrong — it just feels more finished than the underlying thinking deserves.
Has anyone noticed this in their own work, where the polish itself becomes the problem?
