Shubhorup Biswas comments on [missing post]

[Shubhorup Biswas]

Can language models preserve their own alignment?

Wouldn’t consider this an argument for; rather a project proposal to empirically test how much models remain “good”