Comment by pelican0

Is there a clear definition of what Alignment is in OpenAI's perspective, and what the model user can expect of it?

It's one thing if to them it means "it will do what you want following your intentions to the best of its abilities" vs "we will not let you do something dangerous with it unless you're one of us, and that's it".