When a candidate scores low on learning ability, only 34% of them pass Holland Casino’s croupier training. When they score high, that number jumps to 70%. Holland Casino didn’t set out to measure “AI-readiness.” They were just trying to figure out who’d make it through training. It turned out learning ability was the answer all along — and that same trait is exactly what most companies are ignoring right now while they chase something else entirely.
In an era where AI changes what jobs require faster than any hiring process can follow, two things have become true at the same time: CVs are increasingly useless, and everyone wants to hire people who are “future-proof.” Or, described more fancy, “AI-ready.”
A new trend in hiring assessments: AI-readiness tests
A new category of hiring assessments has emerged over the last 12 months: AI-readiness assessments. Can a candidate evaluate AI output? Can they prompt effectively?
It’s understandable why organisations think they need this. But AI-readiness assessments face a fundamental challenge shared with other hard-skill assessments: they measure what someone knows today, not how quickly they can learn what comes next.
Prompt engineering went from an early-adopter specialisation in 2023 to something AI systems handle themselves by 2026. That cycle took two years. The person you hire for their AI knowledge today will most likely need to learn something completely new within two years.
This isn’t a hypothetical concern. According to the World Economic Forum’s Future of Jobs Report 2025, employers expect 39% of workers’ core skills to change by 2030. Assessing current AI knowledge tells you who learned the right things last year. It tells you nothing about who will figure out the right things next year.
The one metric that actually predicts future performance
Organisational psychology has spent over a century studying what predicts sustained job performance. While the world of work has changed drastically, the one variable that keeps predicting it has stayed exactly the same: cognitive ability, not current skill level.
Skills describe what someone has already learned. Cognition describes how fast and how well they can learn the next thing. In a stable environment, that distinction barely matters. In an environment where AI rewrites the rules of work every couple of years, it’s the only distinction that does.
Three cognitive traits matter most here — with the caveat that role complexity always shapes how much each one weighs.
Learning ability
Learning ability is the speed and accuracy with which someone acquires new knowledge and applies it in practice — the capacity to pick up something unfamiliar and make it useful, quickly.
This is exactly what Holland Casino found when building its hiring process for croupiers, a role with a mandatory three-month training programme. Candidates who scored high on learning ability passed training at more than double the rate of those who scored low — 70% versus 34%. Every few months there are new tools, new workflows, new expectations of baseline competence. An employee with high learning ability closes that gap in weeks. One with low learning ability needs months, if not years.
VodafoneZiggo found something similar by accident: learning ability wasn’t even part of their original hiring criteria for customer-facing roles. Once they started measuring it, it turned out to be the strongest predictor of who’d succeed — stronger than anything already in their competency framework.
Cognitive flexibility
Cognitive flexibility is the ability to quickly and effectively adapt to changing circumstances.
Most professionals develop mental models that worked well for years and defend them long past the point of usefulness. In an AI-driven environment, those models get invalidated constantly. The way a marketer thought about content two years ago is structurally different from what the job requires today. Cognitive flexibility determines whether someone updates their approach when the ground shifts, or keeps applying yesterday’s logic to tomorrow’s problems.
Hiring for high cognitive flexibility doesn’t mean hiring someone without conviction. It means hiring someone whose convictions respond to evidence rather than defend against it.
Problem-solving ability
Problem-solving ability isn’t about knowing the right answer. It’s the capacity to find a workable one, often under pressure, with incomplete information, in situations you haven’t encountered before.
That’s exactly the condition AI creates. AI generates outputs constantly — some excellent, some plausible but wrong. The person who can’t independently evaluate what they’re looking at, question its assumptions, and construct a better path when needed isn’t using AI as leverage. They’re sailing blindly on it, until the moment they no longer understand the output.
Problem-solving ability is what separates who effectively works with AI from someone who simply relies on AI.
Why traditional cognitive tests won’t help you here
There’s an important caveat. Many traditional cognitive assessments face the same problem AI creates elsewhere: if a test asks candidates to solve a logical sequence or answer a verbal-reasoning question in multiple-choice format, candidates can ask AI to do exactly that for them. In the blink of an eye, this has made traditional cognitive assessments unreliable for a large share of candidates.
This is a design problem, not just a knowledge-testing problem: a multiple-choice question about a hypothetical scenario can be answered by ChatGPT in seconds. Measuring learning ability, cognitive flexibility, and problem-solving ability reliably means watching how someone actually behaves under pressure — not what they say they’d do. Here’s how to effectively AI-proof your hiring process.
At Equalture, this is why we build game-based assessments instead of multiple-choice cognitive tests. Instead of asking candidates what they think or know, our neuroscience-based games observe how someone actually thinks in real time — through decisions, patterns, and behaviour that can’t be outsourced to a language model.
What this means for high-volume, frontline hiring
This matters most exactly where the stakes of a wrong hire are highest: high-volume, frontline roles, where there’s no time to discover months into the job that someone can’t adapt. Holland Casino’s croupiers had a fixed, expensive training window — 34% versus 70% isn’t a rounding error, it’s the difference between a training budget that pays off and one that doesn’t. VodafoneZiggo’s customer-facing teams face constantly shifting scripts, systems and expectations — exactly the kind of environment where cognitive flexibility and learning ability matter more than whatever the candidate already happens to know.
The organisations that will hire well through this period of change aren’t the ones building sharper AI-readiness checklists. They’re the ones who understood that in a world where core skills keep shifting, the only durable hiring signal is how someone thinks, adapts, and learns — not what they currently know.
Cognition compounds. Skill snapshots expire.
