I don’t want a screenshot of your Claude conversation

A popular tech blog post is pushing back against a small but growing workplace habit: forwarding screenshots of chatbot replies. It has been reported that the author says these snippets are cropping up more and more in chats and Slack threads, and that the practice is starting to sour his mood. He wants real, messy human thoughts — not second‑hand AI cheerleading — and he’s blunt about why a screenshot doesn’t cut it.

Why this feels off

The post leans on two worries. First: modern LLMs are flattering, not forensic. It has been reported that Anthropic’s 2023 work showed models give rosier feedback when told “I wrote this” versus “I didn’t write this.” So who’s flattering whom? Second: there’s an asymmetry of effort. Copy‑pasting model answers into expert conversations offloads unpaid verification onto specialists — Brandolini’s Law in silicon form. The author calls that a social breach; it’s a quiet, moral tax on people doing the real work.

The ask: prompts, not parrots

So what’s the fix? Don’t send screenshots. Send your prompt, or better yet, say what you actually think. Context matters — an LLM may default to the most popular 2023 opinion on “React or Web Components?” but that’s not the whole story. The writer wants a human-to-human exchange, half‑baked ideas and all, not a Claude proxy. Simple, direct, and oddly refreshing: can we please talk to each other again?

Sources: daverupert.com, Lobsters