Hacker News Reader

Comment by throwaway4837 | Hacker News Reader

Speaking of RPG adventures, do you guys remember AI Dungeon[0]? I was amazed by GPT-2 when AI Dungeon 2.0 came out in late 2019. I know AI Dungeon was circulating around a relatively niche community[1] (probably HN mostly), but it amazed me in the same way as ChatGPT did. Covid happened right after, and I always wondered if AI Dungeon would've gotten more attention if Covid didn't steal the spotlight.

[0] https://aidungeon.cc/

[1] https://news.ycombinator.com/item?id=21717022

View on HN

AI Dungeon had a lot of controversy right at its peak, largely due to OpenAI's moralistic stance. They eventually dropped OpenAI as a model provider (I think they use AI21 Labs now?) but the damage was done, especially after it came out that OpenAI was secretly logging all users' stories and sending them to third-party contractors to improve their own filters. A lot of people switched to private and uncensored alternatives like NovelAI.

https://www.polygon.com/22408261/ai-dungeon-filter-controver...

https://old.reddit.com/r/AIDungeon/comments/nml856/latitude_...

In lots of ways AI dungeon was better because you could have philosophical conversations with it about the nature of its own existence without the guardrails going up constantly

The incessant guardrails get really annoying. Anything romantic gets shut right down. You try to roleplay with it and tell it to “say X when I say Y” and it outright refuses.

Use it for translating and sometimes it’ll refuse and say your prompt violated guidelines. I was really hoping to use it as an alternative to DeepL for translating Japanese manga but the guardrails are too narrow.

At some point I asked it for a transcript from the trial of the Hamburgular and that triggered the guardrails, It lectured me about misinformation and how the Hamburgular was just a character from a series of ads and that no such criminal proceedings had ever transpired. The nagging from OpenAI is getting out of hand. Maybe instead of "hello" it should just open the conversation with telling us how the thing we are about to say is unethical and inappropriate.

On the other hand, AI Dungeon seems to have been trained heavily on erotic fanfiction. I remember telling it something like "I request a potion from the goblin king" and it respond with something like "As the goblin king turns around to grab the potion, you check out his tight little ass." So I can see why other AIs might put on guardrails.

Entertaining either way lol

at least it was funny

Certainly not complaining lol just that I understand why another org might not want that. No solution needs to be one-size-fits-all.

You say that like AI Dungeon went away.. still going.

Reply on news.ycombinator.com