It makes me laugh anyone cares about a censor. I guess 10 per cent of the userbase wanted to nsfw chat with waifus? Judging by the stable diffusion community on surprised it's not 90% 🤣
I would be interested using it for my work, so offline commercially licenced models are more my interest due to privacy. Unfortunately my 1080ti doesn't seem very capable of running anything useful. (Also the offline models aren't as fast or advanced)
I'll stick to redacting the information I give gpt!
Two issues I ran into were I like midjourney for grimdark artwork, but it shies away from certain things like blood, and getting chatgpt to generate poetry/lyrics with dark motifs.
But more generally, I honestly just hate being mollycoddled by a god damn robot. I'm a big boy, I know what I asked for, and I can handle the results.
Tech/Programming communities have also noticed that recent changes to the censorship model has messed with the validity and quality of the responses ChatGPT is providing.
I've ran into the same issue recently, where some questions I've asked (like converting a JSON to a Typescript type) are now telling me how to do something instead of just doing it. So I've been relying on ChatGPT a little less - and I imagine I'm not the only one.
Actually, I've had a slightly opposite experience. I found that while asking it general programming questions, it initially tried to exain to me how I could achieve what I was looking to do, and lately it has been jumping straight to writing example code (sometimes even asking for my existing code so that it can modify it)
I've definitely found the playground to be more useful than chat, but it of course can cost more if you're burning through a lot of tokens (I've been at least doubling my plus subscription every month!)