anatomyofai
Region: US
Friday 24 July 2026 12:04:47 GMT
1320
103
9
5
Music
Download
Comments
jamesmastersart :
I am a neurospicy writer and my most recent Substack post said it was 59% AI Generated, 0% AI Assisted, and 0% Human. I thought about going in and trying to figure out why, but I didn’t want to spend anymore time on it. Also, I’ve used the words delve and intrinsic since high school. LOL 😆 I also think that topic makes a difference with the new system. I cyber stalked a few of my colleagues. Most of their work was coming up 100% AI generated. Except these are people I know personally. We have conversations about the things we write about. It’s a niche topic, New Thought Metaphysics. The problem with New Thought is it frequently uses broad and inclusive language, because the system of metaphysics focuses primarily on universal spiritual principles, parapsychology, and philosophy. I think that the topic itself might be triggering false positives. I tend to deviate slightly from that specific topic in my writing, which is probably why I had 49%. I also noticed that if any of my colleagues wrote about another topic, it tracked it as 100% human. Final thought, it just depends on what Substack wants to become. It had the potential to be a marketplace of ideas. That said, there are a couple English as a second language people that I follow. Their work is coming up as AI generated too. They have been open about using AI and it’s a phenomenal accessibility tool. Yet, those people will be dinged because English is not their first language. So, it becomes a marketplace primarily for native English speakers.
2026-08-02 08:00:51
1
OSP 🌶️ :
it’s scoring the posts to what end?
2026-07-24 19:32:35
1
Natalya :
This is a massively good thing imo. It would be helpful to get transparency into how the model works but it’s tiring getting recommended yet another substack essay only for it to be painfully obviously AI slop. I was hoping they’d implement anti-AI controls soon.
2026-07-24 14:11:17
1
EB :
this is my whole semester. i run things through walter writes ai detector first now, the reads line up better with what i expect. less noisy than pangram was for me
2026-09-09 07:35:18
0
cuco :
does proofademic consider context like assignment type or discipline
2026-07-27 03:20:36
0
Dr. Jawn Lam :
What a waste. Wouldn't it be more beneficial for them to spend their resources on trying to detect non-human engagement?
2026-07-29 14:58:38
0
luz.9 :
I do not like it 😕
2026-07-24 22:36:02
0
To see more videos from user @anatomyofai, please go to the Tikwm
homepage.
