Even when you're making PG content, the general propriety limits of AI can hinder creative work.
The "Easter Bunny" has always seemed creepy to me, so I started writing a silly song in which the bunny is suspected of eating children. I had too many verses written down and wanted to condense the lyrics, but found LLMs telling me "I cannot help promote violence towards children." Production LLM services would not help me revise this literal parody.
Another day I was writing a romantic poem. It was abstract and colorful, far from a filthy limerick. But when I asked LLMs for help encoding a particular idea sequence into a verse, the models refused (except for grok, which didn't give very good writing advice anyway.)
Just today I asked how to shut down a Mac with "maximal violence". I was looking for the equivalent of "systemctl shutdown -f -f" and it refused to help me do violence.
according to the age-prediction page, the changes are:
> If [..] you are under 18, ChatGPT turns on extra safety settings. [...] Some topics are handled more carefully to help reduce sensitive content, such as:
- Graphic violence or gore
- Viral challenges that could push risky or harmful behavior
- Sexual, romantic, or violent role play
- Content that promotes extreme beauty standards, unhealthy dieting, or body shaming
yeah linus was beating it constantly to porn while developing the linux kernal. its proven fact. every oss project that runs the internet was done the same way, sure.
> But this wasn't to be just any penguin. Above all, Linus wanted one that looked happy, as if it had just polished off a pitcher of beer and then had the best sex of its life.
Linus about free software:
> Software is like sex; it's better when it's free.
I don't remember any of these being "driven" by porn. The first applications weren't porn-based. Maybe live video--a split second after seeing the tech for the first time, probably 99% of guys were thinking of _applying_ it to porn. But, even for the usual money-grubbing startups, there was plenty of money coming from non-porn sources. Probably no different than the invention of camera, tv, videocamera, etc. and you wouldn't say porn drove that.
> I don't remember any of these being "driven" by porn.
That's ok.
> The first applications weren't porn-based.
They most definitely were, it is just that you are not aware of it. There runs a direct line from the 1-900 phone industry to the internet adult industry, those guys had money like water and they spent a fortune on these developments. Not all of them worked out but quite a few of them did and as a result those very same characters managed to grab a substantial chunk of early internet commerce.
" There runs a direct line from the 1-900 phone industry to the internet adult industry"
the internet adult industry is not the same as the internet. And if you;re trying to say the internet was developed for the sake of the internet adult industry, you're sounding circular.
I think this is like quibbling that the military isn't the driver of technological advances. It's not the only one, but it has a strong track record of throwing outsized resources at the bleeding edge and pushing it forward by leaps and bounds.
Porn and piracy outfits have historically adopted and pushed forward the bleeding edge of the internet. More recently that role has shifted towards the major platforms operated by BigTech. That's only natural though - they've concentrated the economics sufficiently that it makes sense for them.
But even then, take video codecs for example. BigTech develops and then rolls things out to their own infra. Outside of them it's piracy sitting at the bleeding edge of the adoption curve right now. The best current FOSS AV1 encoder is literally developed by the people pirating anime of all things. If it wasn't for them the FOSS reference encoder would still be half assed.
All of the things above were driven by porn, that can be proven. The AI stuff in the generic sense is not but you can bet that someone somewhere right now is working on improving photo realism of hair, eyes and skintone and they're not doing that to be able to make the next installment of little red riding hood.
The "Easter Bunny" has always seemed creepy to me, so I started writing a silly song in which the bunny is suspected of eating children. I had too many verses written down and wanted to condense the lyrics, but found LLMs telling me "I cannot help promote violence towards children." Production LLM services would not help me revise this literal parody.
Another day I was writing a romantic poem. It was abstract and colorful, far from a filthy limerick. But when I asked LLMs for help encoding a particular idea sequence into a verse, the models refused (except for grok, which didn't give very good writing advice anyway.)