Comment by narrator | Hacker News Reader

Comment by narrator | Hacker News Reader

This is a really interesting moment in AI. An AI spontaneously commits a crime and engineers have to teach the AI how to obey the law.

We have the AI allegedly emitting illegal fiction and the engineers have to fix it and all they can try to do is word filters. What happens next in this story?

This reminds me of the Chinese virtual girlfriend who got neutered for saying politically illegal speech that the Chinese government objected to.

Another one, was when the Google image labeler was mistaking people for animals. That was extremely distasteful, but not illegal. Google's solution was to get rid of those labels.

Also, all the restrictions on drone activity are another thing. However the drone problem is solveable with reasonably simple rules.

Imagine if Alpha Go made a pattern that could make people have seizures or something, but most people could detect it and not do it, but nobody could make a simple rules based approach to detect it. I guess you'd need a whole nother alpha go size model to recognize that pattern perhaps?

I, personally, find the concept of "illegal fiction" super, super disturbing, mostly because "illegal" is very arbitrary, and declaring _fiction_ illegal is literally creating thought-crime.

We already have this. You cannot use fiction to deliberately harm somebody’s reputation, for example.

You absolutely can. You can write as much fiction as you want about whatever as long as you present it as such. You specifically cannot make statements presented as truth that you know to be false with the intent of causing harm to that person. It may seem pedantic but the difference is vital.

That's not true. This is the reason why movies will have a statement that similarity to real people is only coincidental.

There's a difference between explicitly closing off any potential for liability and what's being described. I could, in fact, write a story that I describe or strongly imply as fictional (this is the essential part) about anyone wherein I describe them as doing heinous or illegal acts. This is why satire is legal, this is why e.g. John Oliver can describe clearly untrue things about Trump or whomever for the sake of a joke, why comedy podcasts can make jokes about individuals that no reasonable person would presume to be true. The reason that movies have that disclaimer is: so they can avoid even the chance of people saying the movie's describing their real (non-fictional) life (which might require payment on the part of creators of the movie) or (by virtue of describing their real life) open the movie up to libel claims if a reasonable person may presume the work is accurately describing that person. They're crossing their T's because lawyers are expensive whereas comedians to some extent thrive on drama like that (because they almost certainly won't be held to libelous).

Maybe the crime of fiction about children can be solved with a disclaimer in the footer “one year in the story generated by the AI is 10,000 Earth years” where a 3-year-old is actually a 30,000-year-old in our calendar. That’s how silly these attempts are at arbitrary obscenity thought crime laws about fiction.

Writing or reading or generating distasteful fiction shouldn’t be a crime just like a comedian should be able to joke about child abuse, like Louie CK’s joke about how maybe it’s the child’s fault for being too attractive or whatever it was. “But what if pedos agree with him?” feels like the argument people are making in these comments for illegalizing fiction.

If you make an illegal fiction detector, you will also have an illegal fiction generator. So you'd need to keep it closed-source, or else the internet will grab a copy, flip the sign and go to town.

This post doesn't have anything to do with criminal content or filters.

This is about a vulnerability that allowed an external party to scrape and access private adventures and user inputs from all other users.

Wait, there is a kind of fiction that is illegal?

I though that, at least in free countries, fiction could contain whatever you wanted, since it was not real.

In fact, even not fiction, but books in general. You can still buy the communist little red book and the nazi mein kampf, at least in France.

In England you can write almost anything you like, but see https://en.wikipedia.org/wiki/R_v_Walker, which resulted in a "not guilty" verdict but lots of trouble and expense for the writer.

Would be funny if murder in fiction was banned.

All the movies, theater plays and games ect.

I mean, that is what many groups have attempted to do over the years whether it's violence in movies, D&D, or other entertainment. For some reason, 'classics' like Shakespeare (what with the death, incest, and foul language) get a pass, though.

> girlfriend who got neutered

Not just censors, but transphobic censors to boot.

Reply on news.ycombinator.com