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 …
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.