The tool is neat, went back through 2 months of my most recent posts and everything was 99.9% human except two posts that were 40-60% human and those posts were only 1-2 sentences long. Since this made me curious I told chatgpt to: "Write a paragraph defending myself that my posts are not created by an AI". This was its response:
"As an individual, I am proud to affirm that my posts are a genuine reflection of my thoughts, opinions, and unique personality. Each word, sentence, and idea presented in my content stems from my personal experiences, emotions, and knowledge. The depth and complexity of my posts are a testament to the authenticity of my human nature. I invest time and effort in crafting each message, carefully selecting words to express my feelings and perspectives. My posts are imbued with the essence of creativity and individuality, as only a human mind can achieve. I firmly assert that my contributions are not generated by artificial intelligence but are an honest representation of my true self."
Then I ran it through the AI detector and it shows 85.2% probability for AI, and goes up to 90% when adding just the quotations to the paragraph.
Edit: I put it through the system again with my first paragraph included and it now says "This is human text" with it at 48.6% human.