Hacker News Reader

Comment by jasfi | Hacker News Reader

Another problem is AI training on AI generated content. This could lead to degradation of results. So perhaps to make everyone happy a new HTML tag indicating AI content would be useful?

View on HN

> Another problem is AI training on AI generated content. This could lead to degradation of results.

That's only the first step in the degradation. In a worse case scenario, where AI-generated crap is not contained, imagine this: young and still immature humans "learning" their styles of writing, information, and social expectations from content generated by AI that itself trained on AI-generated content─and then more AI trains on content generated by these humans that were trained on AI-generated content. And on, and on, and on...

Luckily, the phenomenon of puberty will make sure that young people absolutely do _not_ want to comply ;).

Ah, but they do want to fit in and speak like their peers, and while a significant part of their socialisation happens online, and as long as that is where they absorb most content, AI will soon infiltrate the circle of online strangers with whom they interact. That's all the more so considering that teenagers are prime targets for advertising networks of dubious ethical standards, who will be the first to deploy (or may already have deployed) well trained and self-correcting "How do you do, fellow kids?" NNs in potentially lucrative communities or social networks.

(I'm not always as doom-and-gloom as it may seem, I'm just in a pessimistic mood today for some reason)

Comply with what? Even in culture we tend to have a limited number of subsets of counter culture. The vast majority of the non-compliant to primary culture fits in one of a few common subcultures. As much as HN users try to exclaim they are islands that stand alone from society, we are all dependant on a functioning society for day to day life.

We should ask questions like "Can countless AI's fracture society into a million sub societies disconnected from the reality around them".

> a new HTML tag indicating AI content would be useful

This has about as much chance of working as the evil bit.

https://en.wikipedia.org/wiki/Evil_bit

It worked for GDPR (regulation would be required).

The people using it to make money don't want it to be easily spotted, tag or metadata marking it will never happen unless required by law

There would indeed need to be a law.

That just makes things easier for the legitimate humans to differentiate themselves as being a cut above.

Reply on news.ycombinator.com