How to write a job description for a role that uses AI

TL;DR: A good AI job description starts with the tasks the person will own and shows where AI fits into each one. List the three skills that matter for most roles, which are checking AI output for errors, knowing when to skip AI, and explaining how they used it. Then cut tool names, years-of-experience requirements, buzzwords, and engineering skills the job doesn’t need, and run your draft through the checklist at the end.

If you’ve posted an AI job description lately asking for “experience with large language models and prompt engineering,” you may have heard crickets. Or you got a pile of applicants who weren’t quite who you needed. Don’t worry, you’re in good company.

That’s because AI is now part of the job well beyond tech. In the US, Indeed Hiring Lab counted 822 distinct job titles with AI in the title in the first quarter of 2026, and 63% of them were outside tech, in fields like sales, HR, and customer support. Those roles don’t need someone who can build AI tools. They need someone who uses them every day and knows how to check what comes out. We call these experts AI-fluent.

The thing is, most job description templates were written before this kind of role existed, so it’s not you, it’s the template. The good news is that a few changes make a big difference. This guide covers four changes you can make from how to describe responsibilities to which skills to leave off of your list. At the end, we’ll share a checklist you can run your draft against.

How to write responsibilities around the work AI touches

Write each responsibility around a task the person will own, then say where AI fits into it. The tools come second, once the work is clear.

A line like “Use AI to improve productivity across workflows” tells a candidate almost nothing. It could describe a marketer, an accountant, or a very ambitious toaster. Candidates can’t picture the job, so they can’t tell whether they’d be good at it.

A clearer responsibility has three parts. Name the task, say what AI does in it, and say what the person checks or decides. Here's how that looks across a few roles:

Marketing manager: “Turn campaign data into a first-pass report with AI tools, then check the numbers and claims before it goes to leadership.” The task is the campaign report, AI’s part is the first draft, and the check is the numbers and claims.
Customer success lead: “Use AI to summarize account history before renewal calls, then confirm key details against the customer record.”
Ops lead: “Build and maintain AI-assisted workflows for vendor intake, and review flagged exceptions by hand.”
Recruiter: “Draft outreach messages and screening questions with AI, then edit them for tone and accuracy before sending.”

Lines like these show candidates where AI fits into the day-to-day. They also make it clear that the person's judgment is part of the job. PowerToFly's Who makes AI work? guide puts it simply: define the work before you define the candidate.

Once the work is clear, picking the skills to list gets much easier.

What AI skills should you list in a job description?

For most roles that use AI, list three skills: checking AI output for errors, knowing when to skip AI, and explaining how they used it. These keep AI-assisted work accurate, and they apply to whatever tools your team happens to use. Basic AI use is becoming as expected as spreadsheet skills in many office roles, so these three focus on the judgment behind it.

Checking AI output for errors

AI tools can sound confident and still be wrong. The person in this role needs to catch a made-up statistic or a summary that drops the key point.

Sample wording: "Reviews AI-generated drafts for accuracy, tone, and missing context before sharing them."

Knowing when to skip AI

Good AI users know where the tools fall short. In a Harvard Business School field experiment with Boston Consulting Group consultants using GPT-4, people who used AI on a task outside its strengths were 19% less likely to reach the correct answer than people who worked without it.

Sample wording: "Knows which tasks benefit from AI and which need a human from start to finish, such as sensitive customer conversations or work involving private data."

Explaining how they used AI

You want someone who can walk a teammate through what they asked the tool, what it got wrong, and what they changed. That habit makes AI-assisted work easier for the rest of the team to review and repeat.

Sample wording: "Documents how AI was used on shared work so teammates can review it and build on it."

Some roles need a more detailed description. For example, a content marketer might need to be a strong prompt writer for the various content-related tasks they’ll have to carry out. Add extras like that only when the role calls for them. If you’re seeing the same gaps across your whole team, our guide to closing the AI skills gap can help you plan for them. Now for the fun part…deciding what to take out.

What to leave off your job description

Leave off anything that shrinks your applicant pool without telling you who can do the job. These three requirements tend to cause trouble.

Tool names the job doesn't depend on

AI tools change fast, and today’s must-have tool is next year’s “oh, we switched.” A strong candidate may also use a different tool that does the same thing. If your team runs on one specific system, say so. Otherwise, describe the work the tools support and let candidates bring what they know.

"Years of AI experience" requirements

The current generation of AI tools is new. OpenAI launched ChatGPT on November 30, 2022, so a requirement for five years of experience with it, or with the tools that followed, asks for more time than they’ve been around.

Katy Peichert, EMEA and LATAM talent partner at Front, has seen this play out:

"It is indeed ridiculous when I see people asking for 10 years of RAG pipeline experience."

Retrieval-augmented generation, or RAG, is a common way of connecting AI models to a company’s own data. Researchers introduced it in 2020, so 10 years of RAG experience isn’t possible. Our guide flags the same problem with AI job titles, specific degrees, and big-name employers, which can screen out people who’d do the job well. A conversation about how candidates use AI in their work today tells you far more than a year count does.

[WHO MAKES AI WORK CTA BANNER]

Engineering skills the role doesn't need

Python, model training, and machine learning frameworks are essential for AI engineers. A marketing manager who uses AI to draft copy won’t be training models between campaign meetings. Listing those skills anyway can scare off the marketers you’re hoping to reach, and it can pull in engineers who expected a very different job. If you do need someone to build AI systems, our guide on how to hire AI developers and engineers covers that search.

With the extras cleared away, the next step is checking whether your post is turning away the people you want.

Why aren't qualified candidates applying to your AI role?

Qualified candidates often skip an AI role when the job description asks for the wrong things, or asks for everything. Vague buzzwords, long wish lists, and requirements no one could realistically meet lead strong people to decide they’re not a fit before they apply.

Watch for these mistakes in your post:

Buzzwords like "AI-first mindset" don't describe a task, so candidates can't tell what you'll ask them to do.
Long wish lists can push people out. LinkedIn's 2019 gender insights research found that women apply to 20% fewer jobs than men. Its advice to employers is to rethink long lists of requirements and nice-to-haves, which can deter women who want to check every box.
Mismatched titles confuse people. A marketing manager post that reads like a machine learning job leaves marketers and engineers equally puzzled.

Our guide to writing inclusive job descriptions has more ways to open your post up to a wider pool of strong candidates.

Before and after: fixing one section

Let’s take a look at what this can look like in practice. This is the requirements list from a marketing manager post.

Before:

3+ years of experience with LLMs and prompt engineering
Expert in ChatGPT, Jasper, Midjourney, and Copilot
Familiarity with Python and machine learning a plus
AI-first mindset

After:

Uses AI tools to draft campaign briefs, copy, and reports, and edits the output to match our brand voice
Catches factual errors and off-brand claims in AI drafts before anything goes live
Knows which work should stay fully human, like sensitive customer messaging
Can walk us through how they used AI on a recent project, including what they’d do differently next time

The rewrite asks for the same AI fluency. It just describes it in terms of the work, so a strong marketer can read it and recognize themselves.

A quick checklist for your AI job description

Before you hit post, run your draft through these questions. It’s quick, and it beats a second round of crickets.

Does each responsibility name a task, AI’s part in it, and what the person checks?

Do your skills include checking AI output, knowing when to skip AI, and explaining how AI was used?

Have you removed tool names the job doesn’t depend on?

Is every “years of AI experience” requirement gone?

Do engineering skills appear only if the person will build AI systems?

Have you swapped buzzwords like “AI-first mindset” for real tasks, and trimmed the wish list?

Does the job title match what you’re asking for?

Would a strong candidate, even one new to AI terms, understand every line and recognize themselves in it?

Frequently asked questions about AI job descriptions

How do you write a job description for a role that uses AI?

Start with the tasks the person will own and describe where AI fits into each one, plus what the person checks before the work moves on. List practical AI skills, such as reviewing AI output for errors and knowing when to skip AI. Leave off tool names the job doesn’t depend on, years-of-AI-experience requirements, and engineering skills the role doesn’t need, so strong candidates can see themselves in the job.

What AI skills should I list in a job description?

For most roles, list three skills: checking AI output for errors, knowing when AI isn’t the right tool, and explaining how they used AI so teammates can review the work. Add role-specific skills only when the job calls for them, like prompt writing for a content marketer, where a prompt is the instruction you give an AI tool. These skills keep AI-assisted work accurate, and they apply whatever tools your team uses.

Why aren't qualified candidates applying to my AI role?

Strong candidates often skip AI job descriptions that use vague buzzwords, long wish lists, or experience requirements that outrun how long the tools have existed. When a post doesn’t describe the actual work, people can’t tell if they’re a fit. Rewrite responsibilities around real tasks, cut nonessential requirements, and make sure the job title matches what you’re asking for.

Find AI-fluent experts for roles across your business

A clear job description gets the right people interested. PowerToFly can help you meet them, with a community of 380K+ verified experts and 65K+ AI specialists, available full-time, on contract, or project-to-hire. For more on building the right team around AI, see our guide to hiring the right AI team.

Talk to PowerToFly about finding AI-fluent experts for roles across your business.