The Rise and Plateau of the Prompt Engineer

It was the hottest job title of 2023. Here's what actually happened to it.

A Premium That Didn't Last

Three years ago, "prompt engineer" was everywhere.

Job postings. LinkedIn headlines. Bootcamp curriculums. The pitch was simple: AI is powerful but difficult to use well, so the people who know how to use it will command a premium. Six-figure salaries for people who knew how to write the right instructions to an AI model.

It was not entirely wrong. But it was significantly overstated. And what has happened since tells you something important about how AI skill sets actually evolve and what that means for your career.

What is Prompt Engineering?

Start with the basics. A prompt is simply the instruction you give an AI model or the text you type into ChatGPT, Claude, or any other AI tool to tell it what you want. "Summarize this document." "Write me an email declining this meeting." "Explain this concept like I'm ten." That's a prompt.

Prompting matters. The way you frame a question, provide context, structure your instructions, and iterate on outputs genuinely affects the quality of what AI produces. That is not nothing.

But here is the problem with treating it as a standalone profession: the models keep getting better at understanding intent. What required careful, structured prompting in early AI models increasingly just requires clear communication in current ones. The skill floor has risen dramatically. What was a differentiator in 2023 is now table stakes.

The people who built careers exclusively on "I know how to prompt AI" are finding that the moat dried up faster than expected. The professionals thriving today treat prompting as just one tool in a deeper skillset. They succeed because they anchor that tool in domain expertise, product thinking, security knowledge, or engineering ability.

What the Market Actually Rewards

The job titles that are genuinely commanding a premium right now are not "prompt engineer" anymore. They are roles where AI proficiency is layered on top of irreplaceable domain expertise:

AI Security Engineer - someone who understands both how AI systems work and how they can be attacked. Demand is significant and supply is extremely limited.

AI Product Manager - someone who can translate genuine user needs into AI-powered product decisions, with enough technical fluency to know what's actually buildable and what's theater.

Forward Deployed Engineer - Companies like Palantir popularized this role, which is now spreading across AI-first startups. Part engineer, part consultant. They embed directly with clients to build and deploy AI systems on-site, in real time. It requires technical depth paired with strong client-facing communication skills.

The pattern is consistent: AI as an accelerator for deep expertise, not a replacement for it.

Safe Harbor: Three Things You Can Do This Week

• Audit your AI skill set honestly. Are you using AI to do things faster, or are you using AI to avoid developing deeper expertise? Both are valid in the short term but only one compounds over time.
• Look at job postings in your field. Search for roles that include AI as a requirement alongside your existing skills. That intersection is where the opportunity lies.
• Pick one area of deeper AI knowledge to develop. Block out some time to map your current domain expertise against an AI framework. If you are in finance, look up how Model Risk Management applies to LLMs. Anchor the technology to what you already know best.

Next week: The cybersecurity divide. Protection isn't equally distributed, and the gap between the secure and the vulnerable is widening fast.