Why good AI matching can make the labour market not less human, but more human.
Last Saturday I was in NRC, one of the leading Dutch newspapers. Wouter van Noort spoke to me about AI matching and the future of job applications. A good reason to take a longer look at a conviction that has only grown stronger in me over the past few years: we should no longer want cover letters.
Not because motivation is unimportant. On the contrary. But because a cover letter shows only very little of who someone is, what someone can do and in what circumstances someone comes into their own. In fact: with a cover letter we mainly measure how well someone can write a cover letter.
We favour people who are good with language, who know the unwritten rules of applying for a job, who have enough time to prepare and who can judge well what an employer likes to hear. And since the arrival of generative AI it has become less and less clear whether a letter says something about the candidate or mostly about the quality of the prompt that was used. That makes the cover letter an increasingly strange selection instrument.
A CV looks back, while work changes forwards
The CV has major limitations too. A CV mainly tells you where someone has worked, which job titles they have held and which qualifications they have completed. It is a summary of the past. But employers want to know what someone can contribute tomorrow.
And it is precisely that question that is becoming more important now that work is changing faster and faster. AI does not take over entire jobs in one go, but it does take over more and more separate tasks. As a result roles change, parts of work disappear and new combinations of human and digital skills emerge. A traditional job title then tells you less and less.
Two people with the same job title may in practice have done very different work. At the same time, people from completely different professions may have surprisingly comparable skills. A logistics planner may have skills that are valuable in healthcare. A recruiter may be a good fit for client advice or project coordination. And someone who for years has combined running a household, informal care and voluntary work may have developed qualities that never become fully visible on any CV.
If we keep looking only at qualifications, job titles and years of experience, we miss a large part of the available talent.
We need to learn to see each other better
The human scale of the labour market has disappeared. There are millions of people, vacancies, projects, courses and possible career steps. No candidate, recruiter, career coach or employer can take in all those possibilities.
On top of that, we humans are quite biased. Often without realising it ourselves. We mainly recognise what we already know. We look at familiar qualifications, employers and career paths. We look for someone who resembles the person who did the work successfully before. A CV that deviates quickly feels like a risk.
That is why the call to simply get to know each other better is likeable, but not enough. The labour market and the amount of available information have simply become too large for that. We need technology to make human possibilities visible again.
This is something other than comparing CVs on keywords
AI matching is still often confused with automatically comparing words from a CV with words from a job advert. That has been going on for years. It is scalable, but hardly intelligent.
Good AI matching goes much further. It starts with a richer picture of people and work. Which skills has someone demonstrably used? Which additional skills have probably been developed through earlier experiences? Which tasks fit with those? Which skills are still missing? What can someone learn relatively easily? And which next steps become reachable as a result?
On the other side, work also needs to be described far more precisely. Not just as one fixed job, but as a combination of tasks, responsibilities, context and required skills. Only then does real matching emerge.
At 8vance we work on this question every day. We use AI to make connections between skills, experiences, vacancies, career steps and development options. Not only to find the most obvious candidate, but precisely also to discover possibilities that people and organisations did not yet see themselves. For me that is the real power of AI matching: not carrying out the same selection faster, but making fundamentally more possibilities visible.
Motivation is not a fixed given
There is another problem in the traditional cover letter. We treat motivation as if it were a fixed characteristic that someone has to be able to prove in advance. But motivation often only arises when someone can genuinely see a possibility.
It is hard for someone to get enthusiastic about a career step they do not know exists. Or about a vacancy that is written in internal jargon and consists mainly of a long list of requirements.
Good matching is therefore not only about the question of whether someone is a perfect fit right now. It is also about questions such as: could this person do this work? Which existing skills are transferable? What would someone still have to learn? Which development becomes possible through this? Does the work connect with what someone finds important? Is there energy when this possibility becomes visible?
So motivation does not always have to be the starting point of the match. It can also be the result of an unexpected but attractive perspective.
AI can see more of someone, but never the whole person
Nuance is important here. AI can process huge amounts of information, recognise patterns and make unexpected connections. But a person is always more than a data profile. Not all relevant information is available. Not every talent lends itself to easy measurement. Personal circumstances, ambitions, insecurities and desires do not fit entirely into an algorithm.
AI should therefore not decide on its own where someone does or does not belong. It should help people to see more possibilities and to have better conversations. Where the data is rich and reliable enough, technology can take care of a large part of the first matching. Where context is missing or not enough trust develops, a human employment adviser, recruiter or career coach remains of great value.
The future is not human or technology. The future is technology that discovers possibilities on a large scale, combined with people who listen, ask further questions, guide and create trust.
Good AI matching asks for good conditions
AI matching is not a magic solution. Its quality stands or falls with the way we design and deploy it. As far as I am concerned, at least four conditions are essential.
Good and up to date data. When profiles are incomplete and jobs are badly described, even the best algorithm cannot make a reliable match.
Explainability. People have to be able to understand why a particular possibility is being suggested. A match score without an explanation is not enough.
Control for the individual. AI should offer possibilities, not dictate a career. People have to be able to add to their profile, indicate preferences and keep making their own choices.
Deliberate checks on bias. Technology can help reduce human bias, but it can also reproduce it when historical data is used uncritically. That calls for continuous testing, transparency and accountability.
Precisely because matching has direct consequences for people, we must not settle for a black box here.
From applying to navigating
My ideal picture is a labour market that looks more like Google Maps. You do not have to know all possible destinations in advance. You see where you are now, which routes are available, which intermediate steps are possible and what you still need to develop along the way.
Perhaps the fastest route is the obvious one. Perhaps an alternative route suits your life better. And sometimes you discover a destination you had never thought of yourself.
That is the shift AI matching makes possible. From searching on job titles to navigating on the basis of skills. From selecting on the past to looking at future potential. From rejecting people because they do not fit completely to making visible what is still needed. From a labour market full of closed doors to an environment in which many more routes become visible.
The real promise is better mobility
If we organise this well, we do not only improve the application process. People can then move more easily to other work. Employers discover talent that currently stays out of sight. Employees gain more insight into internal career options. Jobseekers become less dependent on their ability to sell themselves. And society can connect human talent far better to the places where it is needed.
That asks for more than a new algorithm. It asks for good data, reliable technology, different HR processes and the willingness to question our familiar way of selecting. But the potential is enormous.
Thank you for the interview, Wouter. There is so much more to say about it, but you captured the essence beautifully.
Let us stop judging people mainly on how well they can sell themselves on paper. Let us use AI to really see each other better.