What is a prompt engineer, and is it a real career?

A prompt engineer is someone whose main job is writing and refining the instructions you feed an AI model so it reliably produces useful, accurate output for a specific app or workflow. That means drafting system prompts, building few-shot examples, testing weird edge cases, and rewriting things when the model gets it wrong.

Is it a real career?

Yes and no. The title surged in popularity right after ChatGPT brought AI to a mass audience, and companies rushed to post standalone openings for people who could talk to chatbots. For a while it looked like a brand-new profession built entirely around wording prompts. That has not held up as a job in its own right.

The work did not disappear — it moved inside roles like AI engineer, applied ML engineer, and AI product manager, where writing good prompts is now one task among many rather than the whole job.

Why did the standalone title fade?

Two things drove the shift:

• Models got more forgiving — newer models handle vague, everyday phrasing far better than the earliest ones did, so the gap between a hand-tuned prompt and a normal request has shrunk for most simple tasks.
• The work got bigger — AI products increasingly run as multi-step agents that call tools and files, so the real engineering moved from wording a single prompt to designing the whole system around it.

Should you still learn it?

Absolutely. The skill is more useful and more widely needed than ever — almost anyone building with AI has to do it well. What is rarer is hiring for it as an isolated title.

If you are good at it, expect to use it as part of a broader role, not as your entire job description. Learn to think about how a model interprets instructions, and that skill travels with you into whatever the title happens to be. The wording on the job posting changes, but the underlying ability — getting a model to do exactly what you need, reliably — only gets more valuable as more products are built on top of AI.