I assume his concern is that a proposed solution to avoid losing your job as a SWE is to essentially become a prompt engineer.
More or less, but I guess this was to be expected. Why would making prompts be difficult when LLMs are already capable of fairly difficult programming?