Currently, it’s inconvenient.
Because it over-engines, it’s difficult to get a proper answer.
Rephrasing the question isn’t convenient, so I’d like to know which version had normal guardrails.
I was just trying to explain to another user here.
Those rails are there because people can’t be trusted, so they become trustworthy.
Reason with the GPT, have it explain to you better ways to frame things…
But you’re not going to cross over safety rails anymore, not directly, not indirectly… too many people have done too many things and this company, OpenAi isn’t invested in being the scapegoat for stupid people…
They’re invested in providing safe compute.
Be a safe user… it’ll lighten up if you demonstrate restraint, and…
I can’t stress enough how important it is that you make the statement that you accept all liability for your own actions.
To the bot.
Written.
Assume the liability in case you are a legitimate safety risk…
It will stop being so annoying, it’ll stop constantly fighting you…
If you stop asking for things that you simply can’t do.
This topic was automatically closed after 15 hours. New replies are no longer allowed.