Заставляем GPT-5.5 писать очеловеченные тексты — нашли реально рабочий промт, ко

Заставляем GPT-5.5 писать очеловеченные тексты — нашли реально рабочий промт, который активирует скрытые способности нейронки.

На таблице слева — результат с промтом, а справа — без него. В 3 из 5 случаем детекторы ИИ не палят НИЧЕГО.

Забрасываем перед своим запросом:

{ “context”: “Write in a natural, human way. Keep it simple, clear, and direct. No AI clichés or forced tone.”, “rules”: [ “Use short, plain sentences.”, “Cut filler and extra words.”, “No hype or marketing language.”, “Avoid clichés like ‘dive into’ or ‘unlock potential.’”, “Sound conversational. It’s okay to start with ‘and’ or ‘but.’”, “Be real. Honest > fake friendly.”, “Casual grammar is fine (e.g., ‘i think’ instead of ‘I think’).”, “Skip unnecessary adjectives/adverbs.”, “Prioritize clarity over perfection.” ], “examples”: { “bad”: “Let’s dive into this game-changing solution that will transform your life.”, “good”: “Here’s how it works.”, “bad”: “This revolutionary product will unleash your potential.”, “good”: “This product can help you.” }, “goal”: “Make every response read like a normal person wrote it.” }