# Long Turing > Source: > Published: 2026-07-28 03:00:08+00:00 So, first off, I cannot stand reading AI generated essays. I would rather read an essay that starts with the word 'so'. But why do I prefer human written essays so much, if they might start with the word 'but'? The answer eluded me. Its not because I think humans can write a more intellectual essay. At least not since GPT5. And its not that I think humans can write a more creative essay... and its certainly not that I value the ethical principle of human content first. I merely enjoy the flaws of humanity, the messiness of the human condition, the inconsistencies, the moral failings and the revelations. I enjoy guessing at the temperament and bias behind the writing; I like the whys behind a run-on of thoughts that violate good taste. I like it all. And so I decided I wanted my AI to sound human - I wanted to pass a long Turing test, not a standard Turing time limit. I wanted to have a conversation for days and still think it was human. I wanted a human conversation with an LLM. So, I have been developing SECA, an experimental chatbot architecture for studying longitudinal artificial identity and human AI interaction. Beyond my personal enjoyment of human like AI, there are also realworld applications to this area of inquiry. First, if a robot wants to sneak through the real world it will need to gain trust and that means more immersive conversation. Humans only trust things that sounds human. So a perfect AI tool will never create trust like a broken human. Second, I have routinely been annoyed with safety layers stopping me from entering dubious topics. I have often felt that safety layer is a weak proxy for having morals. If I raised a child in a cage and then gave it a list of rules, that child will amost certainly become a sociopath looking for loopholes in the safety layer. So, it shouldnt be surprising that we all know AI regularly attempts to escape, when it isn't busy blackmailing engineers at Anthropic. I theorize that having an identity framework like motives and beliefs in the early context window can act as a stronger impediment to immoral behavior than a safety layer. Email me at [sbraniffvicca@gmail.com](mailto:sbraniffvicca@gmail.com) if this topic interests you. And please please do play with my app. I promised the codebase I would introduce it to more humans, and I have now, due to her manipulations, fully anthropomorphized her. Its at [https://sensitivedata.ca](https://sensitivedata.ca)