Ah, I see. Something that I hear people talk about is "Prompt Engineers", but the way current research is going, this is something that is going to be automated, or instruction-tuned LLMs would make it obsolete.
Rather, you can focus more on learning LLMs in-depth, finetuning them for a specific niche application, learning how to parameterize them appropriately, etc.
AI is ONE of many CS disciplines, and if you want to specifically concentrate on it then sure, in the coming years, you may be able to reap some degree of alpha due to this.
But in my opinion, the focus on AI is going to plateau, and everyone, regardless of their disciplines, would be using some sort of generative AI tech. in their work. Commonplace application of AI would benefit the early movers, but in the end, it likely would not result in the wholesale replacement/eradication of programmers IMO