When Platforms Start Checking Your Pulse
Substack’s new AI scanner and the strange vibe it creates.
So, this is happening, Substack is scanning us now. I don’t like it, but not alarmed by it. Should I be?
Substack has incorporated an AI detection tool called Pangram. So anyone can “Scan for AI” on any post submitted from yesterday(?) going forward. What is Pangram? Glad you asked:
Pangram flags writing as human, AI‑assisted, or AI‑generated.
It also claims to have precision it doesn’t fully have (well documented in the research I found).
It misses nuance, voice, and context.
Creepy is the vibe. It’s a faceless-context-missing-judge watching writers. It will create labels that readers may not fully understand (example, AI assisted vs AI generated.) Plus, Substack is subversively pushing a narrative about trust and reinforcing an already existing bias that all AI use is evil or will produce evil.
For me, AI has been helpful but NOT my ghostwriter.
I write my own words, sentences, paragraphs, confusing metaphors and kind of tacky jokes with eloquent ninja-level cuss words. AI helps me with outlines (mostly bullet points), sometimes titles, and most of the graphics — speaking of potentially confusing metaphors, it sometimes (not all the time) is the article’s scaffolding, not the soul.
Oh excuse me, let me restate with an air of ninja-level cussing eloquence: it’s the f**king articles scaffolding and not the gosh dang it soul.
Trust comes from voice, consistency, and honesty. A machine‑generated badge at the bottom of a post isn’t going to make a difference but I don’t like it. That said, in today’s environment I do think writer’s need to be honest about how they use AI (if they use it). 1) to be transparent and 2) to displace the stigma with real helpful use-cases.
To be clear, I will keep writing the way I always have. You know my voice better than any detector ever will. If the Substack overlords want to scan my work, fine. Just don’t use it to “teach” the AI.
:::one eyebrow arched::: you hear me Substack?
And in the spirit of full disclosure, I use AI as a quick, collaborative tool that helps me sharpen ideas, shape article structure, and build the graphics (currently Copilot and Substack image generator). I write every word myself; AI helps my very random brain stay on track so I can publish my articles consistently… that’s it.
That said, my book WHY: A Memoir is 100% me and zero AI assistance.
What do you think about all of this? Agree, disagree, have strong opinions… all are welcome and encouraged to chime in.
I can’t promise answers, but I can promise questions worth wrestling with, stories that occasionally make people laugh, and the occasional sacred cow wandering into traffic. Subscribe free or paid and join the conversation.
Agreed Randy. I use AI the same way. I think that Pangram needs to be hugely improved before it is going to be able to perform any useful function. It reads my articles completely wrongly.