One Brother Yells at AI, the Other Asks Nicely — Who Wins?

A Reddit user shared that he and his brother both use LLMs for coding, but they treat them in opposite ways. He has started using a local GLM 5.3 Flash instance — q4 qat — while his brother uses GPT-6-Astra as his daily driver. His brother has repeatedly said his approach is to berate the AI whenever it makes a mistake so that it actually does what he wants, which involves a lot of swearing and "are you an idiot!?" Meanwhile, the user looks at his q4 quant of GLM 5.3 Flash and thinks "aww it's dumb in some ways but it's trying its best, oh it did something!" He is autistic, so he has learned to communicate with precision, which oddly makes talking to small LLMs easier, and he is autistically specific with his requests and asks a lot of questions. He finds it funny to imagine his brother berating a giant model in the other room while he's petting a tiny one. He asks what other people's prompting styles are and what their main LLM is that made them that way.