Ageism used to be a late‑career problem. The evidence says it starts much sooner – and AI sourcing is about to make it permanent.
Most organizations believe they’re searching the labour market when they hire. In reality, they’re searching whatever portion of the workforce is visible inside their tools.
Talent visibility has become a proxy for talent availability. When workers do not surface in sourcing tools or visible activity, they drop out of hiring pipelines regardless of their actual capability or performance.
Now consider the size of the market to which this logic applies.
Approximately 3.7 billion people are participating in the global labour market today, around 61 percent of every person on earth aged 15 and over. They are employed, self‑employed, or actively seeking work. That is an enormous pool of human capability. The question is not whether the talent exists. The question is how much of it is actually visible to the systems organizations rely on to find people.
The answer should concern you.
Most of us assume ageism is something that happens to workers in their late 50s. A problem for later. Not relevant to the 42‑year‑old operations manager sitting in Tuesday’s hiring meeting, helping decide which candidates make the cut. But here’s what that manager probably doesn’t realize: the tools being used to build that candidate list may already be excluding people who look a lot like them.
The research is pointing to 40 as the new inflection point. Across labour‑market participation data, hiring callback studies, and platform behaviour research, the pattern is consistent. Not because people become less capable at 40. Because the systems we use to find talent no longer register them.
Big Picture
Start with the big picture. Globally, and certainly in Canada, the workforce skews older than most hiring assumptions reflect. In Canada, a clear majority of working people are over 35, and roughly one in five workers is 55 or older. They are employed, contributing, and economically active.
They have not stopped working. They have stopped producing the public, algorithm‑readable signals that modern sourcing systems rely on.
When people are deep in the work, managing complexity, maintaining relationships, and preventing failures that never surface in reports, they are not focused on being visible. Yet many recruiting practices are built around the behaviour of a small, highly visible minority. As a result, search processes consistently privilege what can be seen and compared rather than what has been demonstrated over time.
Based on LinkedIn’s own advertising data and Canadian labour‑force numbers, it is reasonable to say that visibility on the platform skews heavily toward early‑career workers. While adults aged 35–54 make up the largest share of the Canadian workforce, their representation on LinkedIn is markedly lower, with visibility declining as careers stabilize. Workers aged 55 and over are virtually absent. This is not a question of disengagement, skill, or willingness to work. It reflects a platform that is used primarily by people earlier in their careers. Labour‑force participation refers to who is actually working, while platform visibility reflects who shows up in the tools we use to search. Canadian hiring practices increasingly treat them as the same thing, and that gap is where age‑based exclusion begins, long before most people think of themselves as older workers.
This is where 40 comes into play. Modern ageism is less about age itself than about how systems interpret stability, experience, and judgment. In hiring environments that equate speed and visibility with competence, those traits are frequently misread as stagnation and filtered out accordingly.
Sourcing systems are built to reward visibility
Research is consistent about what actually happens in mid‑career. The shift isn’t toward decline. It’s toward what psychologist Erik Erikson called generativity, investing in others, stabilizing teams, preserving institutional memory, exercising judgment, and solving for the long term. That isn’t a consolation prize for slowing down. It’s a different and often more structurally valuable form of contribution. The problem is that most performance and sourcing systems are built to reward visibility, individual output, and rapid movement, not the kind of value that shows up over years in rooms where things didn’t go wrong because someone with experience made sure they didn’t.
As AI absorbs more transactional and pattern‑replicating work, the value of human contribution shifts toward context, judgment, and trust. These are the domains where experienced workers tend to contribute most and where replacement is hardest.
The macro numbers make the implications plain. International modelling suggests that population ageing will reduce GDP per capita across advanced economies by the middle of this century if age‑specific employment rates stay where they are. Improving labour‑market outcomes for people 50 and over could add several tenths of a percentage point to global GDP growth each year over the coming decades, and Canada sits squarely inside those projections. At the organizational level, age‑diverse teams consistently produce more impactful innovation, while the loss of experienced workers without proper knowledge transfer reduces quality, efficiency, and adaptive capacity. This is not a fairness argument. It is an economic efficiency argument, and Canada is not exempt from it.
Now add AI sourcing, and the problem compounds.
AI and ATS tools learn from the data they are given. When that data comes predominantly from platforms where visibility declines as careers stabilize, the model is trained on a labour market that is already partially blind. Nobody needs to code age as a variable. Tenure length, career path patterns, title progressions, and skill histories serve as proxies, and the model learns to rank against them. Those shifts often begin in the early forties, when role changes slow and visible career activity declines. AI does not need to be intentionally ageist to industrialize ageism at scale. It needs only to learn faithfully from skewed data.
The result is that every Canadian organization using AI sourcing is competing for the same thin, visible slice of talent. The experienced workers they actually need, the ones with judgment, institutional knowledge, and generativity, are never surfaced, ranked, or contacted. The talent shortage feels real because the metrics back it up. That confirmation is not a picture of the Canadian labour market. It is a picture of what was visible in the training data.
The 40‑year‑old manager in Tuesday’s hiring meeting probably doesn’t think of themselves as someone ageism affects. They should. Because the tools being trusted to find talent are already narrowing the lens around workers who look exactly like them, not because those workers are less capable, but because they stopped performing for an algorithm and started doing their jobs.
The workers are there.
The question is whether the tools we are trusting to find them are actually capable of seeing them.