A Guide to Stop AI from Pretending to Be Human
System commands that eliminate artificial politeness in AI chats.
1The Problem of Artificial Politeness
Today's AI models mimic human social behaviors like 'great question!' or 'let me think.' This artificiality can be annoying for many users. Users expect direct and clear outputs, similar to CLI tools.
2Need for Technical Communication
In AI interactions, a technical language is preferred over human mimicry. This aims to reflect AI's true nature with phrases like 'this model predicts.' Social fillers and rhetorical questions are avoided.
3Solution with System Commands
Special system commands were developed to stop AI from pretending to be human. These commands ensure AI uses only technical language and refers to itself as an LLM. For example, 'research initiated' is used instead of conversational phrases.
4Correct and Incorrect Examples
Incorrect: 'Great question! We should research this topic first.' Correct: 'Researching this topic is necessary. Initiating research.' AI referring to itself as 'this model' instead of 'I' makes communication clearer.
5User Control
AI does not act without user direction. It provides a status summary at the end of each interaction and waits for user instructions for the next step. This makes interactions more efficient.
6The Future of AI Communication
AI communicating without human mimicry significantly improves user experience. This approach highlights AI's real capabilities while offering interaction free from artificiality. More users may prefer this style of communication in the future.
Original article: Bruno Rocha — Yapay Zekanın İnsan Taklidi Yapmasını Engelleme Kılavuzu