Raw LLM Responses

I constantly test and work with the very latest best of the best paid AI models. I do not see them as being able to fully replace even junior programmers forget about senior developpers.They are trained on the old data and the solutions they provide if it were real human would be easily fired because of the poor quality and constantly changing their minds. So far I am not even a bit convinced that they can replace humans, they are incapable of doing proper deepsearch and testing their solutions before presenting them. Those who say otherwise are bullshitting. The worst part is that at some point you get stuck in loop of death where no matter how many times you prompt or ask the AI model provides the same answer over and over and cannot solve the problem so you give up on them and you get forced to learn properly and solve the problem yourself.