… not a robot. CAPTCHA was a gatekeeping system designed by humans for the specific purpose of keeping bots out of places they shouldn’t be and to protect human identities. A system, by the way, that is now a thing of the past due to the sophistication of AI agents.
You could argue that Pangram is simply doing something similar, but the angle is different because this is about our words. Words that Jesus said arise out of the abundance of the heart (Luke 6:45). Words that have the power of life and death (Prov. 18:21). We want those words to be human. Yes and amen. But should we seek a machine to make that determination for us? Whether we want to or not, we’re conceding something by allowing an AI model to test us and tell us whose writing is human and whose isn’t. In the right, good, and understandable yearning for the writing we read to come …
The Substack world was abuzz recently with news that Substack has partnered with Pangram, a system that analyzes text to determine what percentage of that text is “human,” what percentage is “AI,” and what percentage is “AI-assisted.” How is it able to decipher the difference between human writing and AI writing? Here’s how their website explains it:
We train an initial model on a small but diverse dataset of approximately 1 million documents comprised of public and licensed human-written text. The dataset also includes AI-generated text produced by GPT-4 and other frontier language models. The result of training is a neural network capable of reliably predicting whether text was authored by human or AI.
In other words, Pangram is an AI model trained to identify AI written text. What is perhaps most impressive about their system is that it boasts a “near zero false positive rate” - no doubt a great encouragement to those who are concerned about the nice folks on the internet utilizing Pangram to go on a massive witch hunt. In all seriousness, if the claims this company makes for its AI model are true, it really is a remarkably effective tool at verifying the human authenticity of any written text. If Pangram says something you read on Substack was 100% written by a human being, it is almost certainly so.
When I published my first article following the integration on Pangram into Substack earlier this week, I couldn’t resist clicking the “Scan for AI text” option to find out if I was, indeed, human. If you had any concerns up to this point, I have good news for you: it turns out, I am human. Or at least my writing is 100% human. I know it because a robot told me so.
And therein lies the profound, unsettling irony in all of this.
As I watched the Pangram verification pop up on my screen to assure me that my writing was 100% human, what came to my mind were the countless times over the years that I’ve clicked on pictures of cars, buses, and stoplights to prove that I am not a robot. CAPTCHA was a gatekeeping system designed by humans for the specific purpose of keeping bots out of places they shouldn’t be and to protect human identities. A system, by the way, that is now a thing of the past due to the sophistication of AI agents.
You could argue that Pangram is simply doing something similar, but the angle is different because this is about our words. Words that Jesus said arise out of the abundance of the heart (Luke 6:45). Words that have the power of life and death (Prov. 18:21). We want those words to be human. Yes and amen. But should we seek a machine to make that determination for us? Whether we want to or not, we’re conceding something by allowing an AI model to test us and tell us whose writing is human and whose isn’t. In the right, good, and understandable yearning for the writing we read to come from the heart, mind, and soul of a human being, we’re effectively handing our two-factor image bearing authentication over to a robot. Is this the right way to gate-keep our humanity?
I’m not convinced it is. But I’ll tell you what I am convinced of.
I’m convinced that the best writing - the writing that is most compelling, most genuine, most true, most good, most wonderful, and most meaningful - is always, inevitably going to come from women and men who are made in the image and likeness of God. Full stop. The machines are getting “better” each day at firing their mathematical synapses to produce clean, tight, and rhythmic prose. Their slop might be getting less sloppy, but they’re not becoming more human. They’re not growing in the wisdom and knowledge and grace of God. They’re not even capable of wanting to. And I think that will show up in the words “written” by AI or assisted by AI. The kind of writing that reflects our Creator can only be done by ensouled creatures. Call me foolishly optimistic, but I’m simply going to trust that that kind of writing is what will end up rising to the top of my own affections and the affections of other humans.
So I’ll be ignoring the existence of Pangram as I read my Substack articles, thank you very much. I think Substack’s heart is in the right place in bringing this tool in, and I can admire that. I want human writing, too. I’m just not willing to subject my fellow image bearers and their words to a robot’s “human” radar in my pursuit of it.
The Weekly Digest
Premier Thought Every Thursday.
All of our recent essays and podcasts, delivered to you. Free.
Free. Delivered Thursday mornings.