[deelly]

What bothers me a bit: if we can't control/manage ChatGPT which is still "just" a language model and not an intelligent entity, then how do we be able to control/teach next much more sophisticated models?