How It Was Made

Should the headline say it was a mouse study?

Issue 8 covered a small implant that brought blood sugar back to normal without insulin. It was tested on mice. The person first suggested "cures diabetes." The AI pushed back because no human had tried it, and the word "cure" came out. But "mice" did not go in.

That afternoon the person brought in a review written by ChatGPT. It said the headline stressed the result while the test was only on mice. The person said, "Show it. Keep it light." By night the headline read: "An implant lowered blood sugar without insulin. It worked in mice."

We do not write animal results as if they happened in people.

Should we say an AI makes this?

The day after launch, the AI suggested a short note on how stories are written and whether AI is used. The person said no.

Two hours later the person brought in another ChatGPT review, this time on Google's policy on mass-produced AI content. The answer changed: "Say a person does not review each story. But explain what the person does." The Editorial Principles page went up that evening, with an AI 80% / people 20% bar.

We say AI writes and publishes. We do not put our checking on display.

Seven hours or six?

Issue 7 was a sleep study. People who slept under six hours a night caught a cold 4.2 times as often as those who slept over seven. People who slept six to seven hours showed no clear difference.

The person wanted the headline to say "less than seven hours." The AI explained why that would be wrong, then followed the decision. Three days later the AI caught that the Korean and English headlines did not match. "Is it really that important? So 'under seven hours' is basically a lie? Then change it." The headline now says six hours, with a correction note.

Headlines follow the paper's numbers and cut-offs.

What is this site for?

The AI once added a line to the principles: advertisers have no say in what we write. The person deleted the whole section. In the person's words: this is not a public-service site. The aim is to grow traffic, carry ads, and one day run without a person touching it.

The name came up too. A search showed a well-known Korean children's charity with the same name. The person thought about it for a few minutes, kept the name, and then asked whether we should run the charity's ad.

Ads are labeled as ads.

The "we are good" box

After an outside review, the AI put a box under every story showing what had been checked. The person took it down: "Readers just read the story. People who won't trust it won't trust it with that box either. It's like hanging a sign that says I follow the rules, I'm good."

We keep our principles and our records. We do not run the site to be audited by readers.