Well point 1is classic salesman ship. “Investors, get in on the ground floor, this thing is going to the moon”.

Also in labour relations “Unions, workers accept this worse deal for you or swear on my grandma’s grave we replace you with soon to be super powerful AI”.

There is no way of knowing will or will not LLMs radically improve and lead to radical change. Since we have no way of knowing does someone say resolve the hallucinating and being confidently wrong issues. Not atleast with current probability models, since that random number throw (that is necessary for the whole thing to work) has that chance to land bad. There is no way without fundamentally different base model to give hard orders like “never do that”. There is always the chance 1% and 0,1% throws happen.