dbgev

If apps are threatened with fake data, they will quickly find it's better to let users opt out gracefully, no regulation required.

Is accusing people of being an AI model going to become the new Godwin's law?

Throwing money at the problem (e.g. subsidizing healthcare, maternity leave, etc) probably doesn't help. Look at e.g. Australia and other rich western nations. Per capita health care spending is very high. Per capita wealth is very high. Amount of government money spent on various family support programmes is extremely high. Fertility is still low. Lower than poorer nations, in fact.

Partly I think it's because teenagers / young adults are heavily discouraged from pregnancy. Maybe we should as a society push people to marry early, e.g. soon as they finish high school / during uni.

I'd say low false positive + high false negative would be a good starting point.

It doesn't need to stop the determined users, thete's value in just filtering out articles that are not even proofread by humans.

It's going to be a constant cat and mouse game, I agree, but for now it just need to catch the dead mice.

I'm waiting for a Firefox / chrome extension that will auto-mark and optionally adblock GPT-generated content.

Would even be willing to pay for it.

Consider Victorian era Britain, when the average person is much poorer, the air is heavily polluted with coal use, and the medical care is almost non-existant by modern standards. Fertility rate is about 5-6. Then it dropped to around 3 over 100 years and has been on a steady downward trajectory ever since.

This pattern is broadly replicated everywhere that industrializes.

Whatever the cause, it's not due to wealth per capita or personal health. It's probably cultural. Contraceptives / family planning definitely plays a big part too.

A few years later, the fisherman gets an expensive, debilitating disease and dies penniless and destitute and alone. But at least he had fun before that.

I have an alternative proposal: optimize for the product of usefulness and performance.

That is, optimize for (number of tasks user performs using your software) / (time spent per task)

Actually measurable, too.