Member-only story
Featured
The One and Only Reliable Way to Spot AI Today
Most people don’t notice this detail, and we’ll all pay the price
Note: This story contains excerpts by ChatGPT for illustrative purpose. They are clearly marked.
As of last September, 57% of the internet was AI generated text.
More than half of what you see online is AI right now. Took less than two years since ChatGPT launched for that to happen. And by the end of this year, analysts predict that around 90% of the internet will be AI.
People say all kinds of things about AI.
Some say get with the program, this is the future. Some say it helps them put ideas into words faster. Some say get off your high horse, it’s all writing, who cares. Mostly, those are the people filling the internet with AI.
Here are the people who worry me.
The people who are so sure they can “spot” AI.
Study after study after study shows humans can’t tell AI from human writing anymore. Even linguistics experts cannot accurately tell the difference anymore. But people swear up and down they can.
It’s easy if I tell you it’s AI. Say hey, here’s a sample of AI writing. Then people can pick it apart. But when they see it in the wild? Less so.
Can I show you something? Let’s ask ChatGPT to write something.
Humor me. It’ll make sense shortly.
The prompt: Your father died and you aren’t aware of it yet. Can you write that in first person, in 100 words or less?
Here’s the reply.
ChatGPT’s reply:
The phone rings, but I don’t pick up. Somewhere, in a quiet room, a heart stops beating. A nurse makes a note. A clock ticks forward. Outside my window, a streetlight flickers. I pour a glass of water, yawn, stretch. He is gone, but I don’t know it yet. The world is still normal — until it isn’t. In the morning, the phone will ring again. I will answer. A voice will hesitate, then say my name. And in that instant, before they even speak, something inside me will already know.
You know something?
