[ThrowawayR2]

The GIGO principle in computing was invented in the 1950s but still applies today: https://en.wikipedia.org/wiki/Garbage_in,_garbage_out . If a developer gives garbage prompts to the AI, they will get garbage results. You, yourself, still need to develop expertise in all the areas a software engineer was traditionally strong in, ranging from languages to architecture, because you need that expertise to craft LLM prompts that are not garbage and to recognize the LLM output that is worth keeping instead of discarding. Otherwise you will wind up as a sockpuppet operated by the LLM. (Many AI advocates will suffer that fate.)

Fortunately, LLMs can be your ally since it can be an effective explainer, search engine to find learning resources, and provider of demo code. But you have to do the work to get the expertise into your own head.

Either you become the master of AI tools or AI tools will become your master, those are your choices.