AI Required: The Most Meaningless Line in Tech Hiring Right Now
Half of all U.S. tech job postings now mention AI or machine learning skills. That number has nearly doubled in twelve months. If you are hiring in tech right now, you have almost certainly approved a job description with some version of “AI skills required.” If you are looking for work, you have almost certainly read that line and wondered what it actually means.
The answer, in most cases, is that nobody knows. Not the hiring manager. Not the recruiter. And definitely not the candidate who just added “proficient in AI tools” to their resume because it seemed like the thing to do. “AI skills” has become the new “must be a team player.” Everyone writes it. Nobody defines it. And it is quietly wasting months on both sides of the hiring table.
“AI Skills” Means Three Completely Different Jobs Depending on Who You Ask
We work with hiring teams every week, and the range of what “AI skills” means from one company to the next is staggering. For one team, it means the candidate should be comfortable using Copilot or ChatGPT in their daily workflow. For another, it means they should be able to fine-tune a large language model and deploy it in production. For a third, it means something vague about “being AI-forward” that even the hiring manager cannot articulate when pressed.
These are wildly different skill sets. One is a behavioral habit you can pick up in a few weeks. The other is years of specialized technical training. Lumping them under the same two-word requirement does not filter for the right people. It filters for people who are good at mirroring keywords back at job descriptions, which, in a market already flooded with AI-optimized resumes, is exactly the wrong signal to select for.
“I Use It to Write First Drafts” Is Not What You Meant by AI Fluency
From the candidate side, the confusion is just as bad. When every other posting asks for AI skills and nobody explains what that means, people do what they have always done with vague requirements: they fake it until they figure it out. They add “AI/ML” to their LinkedIn headline. They list ChatGPT as a tool. They take a weekend prompt engineering course and call it a certification.
We see this constantly in screening. A candidate lists AI fluency on their resume, and when you ask what they actually do with AI in their daily work, the answer is some version of “I use it to write first drafts of things.” That is not worthless. But it is also not what most hiring managers think they are getting when they screen for AI skills. The gap between what appears on the resume and what shows up in the interview is growing, and it is not because candidates are dishonest. It is because the requirement was never specific enough for honest answers to be possible.
Meanwhile, the genuinely skilled people who know how to evaluate AI-generated code, build production workflows around model outputs, or design systems that keep AI tools from creating more problems than they solve look identical on paper to someone who uses ChatGPT for emails. The vagueness hurts the best candidates most.
Most Roles Need Judgment, Not a Model Training Certificate
Based on what we see across the teams we place into, most roles that list “AI skills” actually need someone who can evaluate what AI produces, not someone who can build an AI system. Can you look at AI-generated code and tell whether it will break under load? Can you catch the subtle architectural assumptions a model makes that do not match your system’s actual constraints? Can you fold AI tools into a real production workflow without creating a mess that someone else has to clean up?
That is judgment. It is built through experience, not through a certification. And it is the skill that matters in the vast majority of engineering roles right now. Only a small percentage of companies actually need someone who can fine-tune or deploy models. The rest need engineers who can work confidently alongside AI and know when to trust its output and when to throw it away.
But the job description does not say that. It says “AI skills required.” So companies get hundreds of applications from people who learned to prompt and five from people who learned to evaluate. The role stays open for months. Everyone blames the talent market.
“Review and Ship AI-Generated Code Daily” Is a Job Requirement. “AI Skills” Is Not.
The fix is not to stop asking for AI capability. AI fluency is genuinely important, and it is only going to get more important. The fix is to say what you actually mean. Replace “AI skills required” with something a candidate can prepare for and an interviewer can test for: “You will review and ship AI-generated code daily.” Or: “You will integrate LLM APIs into a production application.” Or: “You will use AI tools to accelerate your existing work and you should be able to show us how.”
Specific requirements attract specific people. Vague requirements attract everyone, which in a market with this much noise means they attract no one useful. The companies we work with that have the shortest time-to-fill on AI-adjacent roles are the ones that stopped writing “AI skills” and started describing the work.
The Takeaway: Your Job Description Is Your First Filter. Make It Work.
The AI talent gap is real. Only about 16% of workers have what researchers would call high AI readiness right now. But the gap is not going to close by adding “AI skills” to every posting and hoping the right person shows up. It closes when companies get specific about what they need, when candidates can actually tell whether they qualify, and when both sides stop wasting three months talking past each other because nobody defined the requirement.
TechX works on both sides of this. We train engineers on real projects so their AI fluency is built on production work, not tutorials. And we help hiring teams define what AI capability actually looks like for their specific roles, so they stop screening for a keyword and start screening for the skill. If your AI postings are attracting the wrong people, we should talk.