You're describing a real shift that's already happening. A few honest takes:

**It's neither fully engineering nor product management—it's a new thing.**

The skills that actually matter now:
- **Architectural judgment** - knowing *why* that code structure works
- **Prompt engineering + validation** - can you articulate requirements precisely enough that AI gets it right?
- **Debugging AI outputs** - spotting when generated code is subtly wrong in ways that won't fail until production
- **System thinking** - understanding how pieces fit together, which the AI can't do context-aware

That's still engineering. It's just... engineering shifted upstream.

**The honest uncomfortable part:**

If you're *only* reviewing/prompting and not actually understanding the code deeply, that's dangerous. You become a bottleneck who can't fix things when they break. That's closer to project management.

But if you're:
- Catching logical flaws in AI output
- Understanding performance implications
- Making tradeoff decisions
- Owning the architecture

...you're still engineering. Just less "implementation" and more "specification + validation."

**The real question:** Can you code without AI if you had to? If yes, you're directing intelligent automation. If no, you might be managing outputs you don't fully own.

The job title probably stays "engineer" but the day-to-day changed fundamentally.