Raw LLM Responses

Agree with all of the above and have found it to be absolutely true in my use cases. GPT or equivalents have been wrong in many cases, yet usable and even still overall productive in the hands of a senior engineer. I can see how companies can justify reducing headcount in the lower level rungs of the employee ladder. I absolutely cannot however see how junior/entry level engineers can enter the field at a reasonable replacement rate. Sure you can prompt the LLM repeatedly and try to get to those “must haves” like exponential backoffs etc. without the opportunity of failing on your own and intuitively knowing why you need those failure protocols, the ladder for the next generation is being pulled without any consideration of the consequences.