Your Next Colleague May Have Been Chosen by a Video Game

Imagine applying for a job.
You send your CV. You pass the first screening. Then, instead of the usual email to arrange an interview, a link arrives.

You open it. A game appears on the screen.
You have to make decisions, react to certain stimuli, remember sequences, allocate resources or choose how much to risk.
You play for twenty minutes. Then you close everything.

Meanwhile, someone has gathered information on how you make decisions, on your attention, your risk appetite or how quickly you learn.

And that information could help decide whether you move forward in the selection process.

It is already happening.

They are called game-based assessments, and they represent one of the most interesting endeavours in contemporary recruiting: learning something about a candidate by observing how they behave while doing something.

The difference seems small. In reality, it changes everything.

The problem is that we have become extremely good at interviews

“What is your greatest strength?”
“Tell me about a time when you had to solve a problem.”
“How do you work under pressure?”

These are perfectly legitimate questions. They do, however, have an obvious limitation: we know they are coming.

A quick search for “common job interview questions” turns up thousands of guides, videos, simulators and, today, artificial intelligence tools capable of building an answer to practically any question.

In our guide on how to prepare for a job interview we stressed exactly this point: preparing should mean arriving with a clear idea of which real experiences to draw on to demonstrate your skills, rather than memorising perfect answers.

In 2026 the problem has become even more interesting.

Companies, too, need to be able to distinguish between the ability to describe a skill and the ability to actually use it.
And this is where the game comes into play.

They are watching how you make decisions

A game-based assessment is designed to observe behaviour during a structured experience, rather than to gauge how good someone is at video games.

Today, for example, Pymetrics by Harver uses gamified experiences to analyse attributes related to attention, decision-making, risk appetite, learning, concentration and behaviour.

The final outcome of the game matters less than the choices made to get there.

How quickly a decision is made.
How behaviour changes after a mistake.
How much risk is accepted.
How limited resources are allocated.
What happens when conditions change.

The candidate is not simply saying: “I am good at making decisions under pressure.” They are making decisions.
This is the most compelling promise of game-based assessments.

From CV to behaviour

For decades, selection has been built mainly on what a person has done. A behavioural assessment instead tries to answer a different question: how might this person behave when faced with a problem they have not yet encountered?

This is also why part of recruiting is shifting its focus from professional pedigree alone towards skills, potential and observable capabilities. In 2026, a major review published in the Annual Review of Organizational Psychology and Organizational Behavior identified the evolution of measurement systems as one of the great changes in selection science: from CVs to social media assessments, from multiple-choice to constructed responses, through to the use of artificial intelligence to develop, administer and score tests.

The CV therefore remains an important source, now flanked by tools that seek different information about the candidate.

Can a video game really tell how you will work?

This is the serious question. Here we need to separate the “wow” effect from the research.
A systematic review published in Frontiers in Psychology analysed 34 studies on game-related assessments in personnel selection.

The conclusion is interesting precisely because it is cautious. These tools can be used in selection, while several of the advantages often attributed to them still require stronger evidence.

A good assessment must demonstrate at least four things.

1. It must be reliable.
2. It must genuinely measure what it claims to measure.
3. It must relate to what it aims to predict, for example job performance.
4. It must limit the distortions caused by a candidate’s irrelevant characteristics.

Looking like a video game does not automatically make a test better.

The results are becoming interesting

A study of 11,574 completed assessments analysed a game-based assessment of cognitive ability scored by a machine learning algorithm.

The test showed a correlation of 0.50 with a traditional measure of cognitive ability and a test-retest reliability of 0.68.

The researchers also analysed a second sample of 3,107 candidates to study fairness-related aspects.

One of the more intriguing findings concerns the candidate experience. Among 4,778 people who provided feedback, the system achieved a Net Promoter Score of 58. Many people therefore appeared to view that type of experience favourably compared with the traditional notion of a test.

Even so, more recent research urges caution. A meta-analysis published in 2025 covering 18 studies drawn from 13 peer-reviewed articles found moderate, statistically significant convergence between game-based assessments and traditional self-report personality measures, alongside strong heterogeneity across the studies.

Some games, then, do appear to measure something of interest. Quality depends on the design of the assessment, its validation and the type of construct being observed.

In 2026 the question is also one of governance

Artificial intelligence has become ever more deeply embedded in selection processes, and Europe has decided that certain uses warrant a particularly high level of scrutiny.

The European AI Act classifies as high-risk those systems used in specific areas of employment and recruiting when they materially influence the assessment, selection of, or access to work. Following the amendments approved in 2026, the main obligations for high-risk systems under Annex III will apply from 2 December 2027.

The principle is already clear: when a system helps decide who can access a professional opportunity, data quality, traceability, human oversight, the risk of discrimination and reliability all come into play.

And this completely changes how we ought to look at games, too.

What if I lose because I never play games?

Could age make a difference?
Could familiarity with certain interfaces change the outcome?
Could a disability make it harder to interact with some mechanics?
Might a person not understand why they are playing, and experience the assessment as having little to do with the job?

These questions matter because an assessment must be valid for what it is trying to measure.

The scientific literature on game-related assessments highlights precisely this point: game design can influence the outcome, and construct validity remains one of the areas where more research is needed.

If we are looking for decision-making ability, we must be sure we are measuring decision-making ability and reduce the weight of extraneous factors, such as experience with specific game interfaces.

But the most interesting paradox lies elsewhere

The more technological selection becomes, the greater the need to understand what we are actually measuring.

A CV can be poorly written. An interview answer can be rehearsed. A test can be practised. An algorithm can find correlations that appear significant. A game can record thousands of micro-behaviours. Each of these tools observes one part of the person and produces a signal that must be interpreted.

And this is probably where the problem stops being merely technological. Every selection system embeds a theory, explicit or implicit, about which characteristics make a person suited to a role. When that theory is translated into an algorithm, an assessment or thousands of behavioural micro-signals, it becomes even more important to know who defined it, how it is verified and who retains responsibility for the final decision.

Perhaps the interview of the future will be about doing something

There is something video games understood long before recruiting did.
To find out whether someone can play, they hand them a controller.

It is a simplification, but it contains an interesting insight.
Perhaps part of future selection will increasingly be built around situations in which a person can show what they can do, not just talk about it.

Because the best candidate is not necessarily the one who gives the best answer to the right question. Sometimes it is the one who, faced with something they had not prepared for, simply manages to work out what to do.