By skyforbes I’m a long-term AI user from China. For years, using AI has felt like playing a single-player game. I’ve been isolated in my own workflow without a community to discuss deep, practical usage. GPT just pointed me to this subreddit, so here I am.
My Journey Like many, I started using AI as a "better Google"—just for facts and explanations. Then I moved to the usual roleplay/task prompts, but I found the outputs too stiff. The AI felt boxed in by my constraints.
The Shift: AI as a Colleague, not a Tool The breakthrough happened when I stopped treating AI as a servant and started treating it as a colleague. I explicitly tell it to:
- Disagree with me.
- Challenge my logic.
- Argue back if my ideas are weak.
I’ve since evolved this into a "Manual Multi-Agent Debate" workflow. I often have GPT and Gemini open simultaneously, feeding them the same topic but assigning them different perspectives to debate each other. I just sit back, watch them fight, and curate the best points. To my surprise, the output from these "arguments" is often superior to anything a single sophisticated prompt could produce. It helps immensely with complex planning, long-form writing, and worldbuilding.
My Question to You I feel like I’ve hit a ceiling with this "manual" approach.
- Does anyone else use this "adversarial/debate" workflow?
- Are there frameworks or methods to optimize this "colleague" relationship?
- Where do advanced users hang out to discuss these deeper mental models?
I’m eager to learn from you all.
PS: I don’t speak English. This post was drafted, translated, and refined through multiple rounds of AI verification to ensure clarity. Hope it gets the message across!