Every resume that crosses a hiring manager's desk this year claims they're "AI proficient."
ChatGPT or Claude. Midjourney or Cursor or Perplexity. It's all there, usually highlighted in a slightly larger font
But none of it tells you whether someone is actually good at using AI.
Just like listing "Microsoft Word" on a resume in 2004, prompting is an entry-level requirement now. Forbes reported that LinkedIn's Skills on the Rise named AI literacy the most in-demand skill for 2025.
But for hiring managers, that creates a harder question: How do you separate someone who knows how to use AI tools from someone who actually knows how to think with them?
The answer isn't another checklist of platforms, models, or prompting tricks. It's judgment.
The candidates worth paying attention to can look at what AI gives them and decide what's useful, what's wrong, what's generic, what needs another pass, and what should be thrown out entirely.
That's the shift hiring managers need to make in interviews. Stop asking only which tools candidates use, and start asking what they do after the tool hands something back.
Stanford's Teaching Commons built a framework for AI literacy that's worth borrowing for hiring, even though it was written for classrooms, not conference rooms.
It breaks the skill into four overlapping domains:
That framework is a useful gut check for hiring managers, because it draws a hard line between people who can operate a tool and people who understand what the tool is doing.
Someone with real functional literacy doesn't just prompt, but they can explain why a prompt worked, why one model outperformed another on a given task, and where the output is likely to go sideways.
Stanford's guide makes a sharp point about what most AI output actually is by default: For straightforward prompting, a chatbot's response tends to land as an average composite voice aggregated from the internet, scraped from scientific papers, forums, and everything in between.
And that's fine when you want something broadly appealing but is otherwise a problem when you're hiring someone specifically because you need original thinking, a distinct brand voice, or work that doesn't sound like everyone else's AI output.
The candidates worth hiring are the ones who already know the difference.
With a click of a button, anyone can generate ten versions of a headline, ten variations on a campaign concept, ten drafts of a subject line. That ability has never been the skill.
The real skill is knowing which one is actually good, being able to explain why, spotting what still reads as generic, and knowing when to throw the whole batch out and start from scratch. You can call it taste, call it editorial judgment, or simply call it discernment. It’s the thing AI can't fake and can't shortcut, and it's the actual competitive advantage in a world where everyone has access to the same tools.
This is the part most hiring managers skip. They ask what tools a candidate uses and then don’t follow up with what the candidate does after the AI hands something back.
Treating AI skills as one universal checkbox is where a lot of hiring processes go wrong.
"AI skills" means something different depending on the job. Marketing leader Kyle Poyar tracked this firsthand: the number of go-to-market job postings requiring AI skills jumped from roughly 65 in mid-2023 to nearly 1,000 by mid-2025.
When he dug into what those postings actually asked for, the range was enormous. Everything from basic tool familiarity to candidates expected to run AI agents that personalize outreach and mine CRM data, all filed under the same "AI skills" label. That's the exact confusion hiring managers need to design around instead of ignore.
Here's what real AI competency looks like for roles reshaping creative and marketing teams:
Look for people who think in systems, not just prompts. That means workflow automation across platforms, an instinct for data quality, real CRM fluency, and the habit of documenting what they build so it doesn't live only in their head. This role lives or dies on whether the automation actually holds up once someone else has to touch it.
This is one of the fastest-growing "AI-native" roles in marketing right now, and it rewards people who treat AI as infrastructure, not a party trick. Strong candidates can talk through messaging synthesis, positioning work, and how they've translated AI-surfaced insights into an actual campaign a team ran.
Prompt iteration matters less here than editing instinct. Look for SEO and AEO awareness, a sense of audience, consistency with brand voice across a body of work, and (critically) a habit of verifying research rather than trusting the first AI-generated citation.
Visual taste is the whole ballgame. The strongest candidates can direct AI tools instead of accepting whatever the default output looks like, and they can tell you, specifically, when a human hand needs to take over.
Please retire that question. It doesn't separate anyone from anyone else anymore.
Instead, ask questions that force a candidate to reveal how they think, not just what they've tried.
A few that work well across creative and marketing roles:
Kyle Poyar's newsletter Growth Unhinged rounded up the questions GTM leaders at companies like Clay and Zapier actually ask candidates, and a few stand out for how much they reveal in a single line:
These questions have a lot of surface area, and anyone who's actually done the work has an opinion. Anyone who hasn't will stall.
These are the types of questions that expose thinking, and answers that just include tool name-dropping will not survive them.
Employment Attorneys at Fisher Phillips flagged a tiering system that Zapier built internally to sort AI capability, and it translates cleanly to creative and marketing hires.
Here are the four tiers, adapted slightly:
Most junior hires will land in that second or third tier, and that's fine. What matters is that a hiring manager can name where a candidate sits and isn't mistaking tier-two tool fluency for tier-four judgment just because the portfolio looks polished.
Candidates overstate AI experience more often than hiring managers might assume, which is exactly why concrete, story-based questions matter more than a checklist of tool names on a resume.
That overstatement problem is also why it helps to know the specific tells to watch for once the interview is actually underway.
Red flags:
Green flags:
The strongest AI-literate hires aren't the ones who can name every model that shipped this month. They bring the same sharp judgment to the work no matter which tool they're using.
That shows up as traditional design thinking, editorial instinct, systems thinking, verification habits, and the willingness to say, "This isn't good enough yet" even when AI handed them something that looked finished. Those skills were valuable before generative AI existed, and they'll outlast whatever tool is dominating LinkedIn headlines a year from now.
Hiring for AI fluency (done well) isn't really about AI at all. It's about finding people with good judgment who happen to be working in an AI-shaped world. That's a search Artisan Talent has been running long before "Prompt Engineer" was a job title.