When discussing artificial intelligence and the labour market, the most common question remains:
“Will AI eliminate my job?”
It is a legitimate question.
But it may focus too heavily on the most extreme possible outcome.
A profession does not need to disappear to be profoundly affected.
The position may continue to exist.
The company may continue hiring.
The professional may remain employed.
And yet the economic value of the work may change.
Artificial intelligence may first appear through lower pay, weaker raises, more precarious contracts, smaller teams, higher targets, and reduced bargaining power.
The job remains.
But the relationship between what the professional delivers and what they can receive begins to be rewritten.
The more useful question may therefore not be only:
“Will my job continue to exist?”
It may be:
“How much will the market be willing to pay for my work once part of it becomes easier, faster, or cheaper?”
Employment and wages do not move together
The number of jobs can remain relatively stable while wages change.
A company may continue to need developers, analysts, and support professionals.
But when tools allow more people to perform part of those activities, the organization can expand the number of candidates considered qualified.
When the supply of professionals grows faster than demand, bargaining power declines.
The company does not need to eliminate the role.
It can simply pay less for the same title.
It may also hire less experienced professionals for work that previously required greater specialization.
Combine two functions.
Outsource the activity.
Replace stable employment with temporary work or independent contracting.
It can maintain nominal pay while significantly increasing expected output.
In that case, the professional does not receive less money.
They receive less for each unit of responsibility, effort, or value produced.
Wage pressure does not always appear as an explicit pay cut.
Sometimes, it appears as the absence of a raise.
AI can democratize a task
Consider an activity that once required several years of training.
Producing an initial version of code.
Creating an analysis.
Structuring a report.
Developing a presentation.
Writing a query.
Interpreting a technical document.
When artificial intelligence reduces the initial difficulty of those tasks, more people can perform them.
This can be positive.
Knowledge becomes more accessible.
Professionals gain autonomy.
Smaller companies can perform work that was previously beyond their reach.
But there is also an economic consequence.
When a skill stops being scarce, its price tends to face pressure.
PwC’s 2026 Global AI Jobs Barometer describes a two-track labour market.
In some occupations, AI acts as a force multiplier for experts. The technology absorbs routine work while increasing the value of human judgment, creativity, and leadership. PwC describes these roles as “professionalised.”
In others, AI makes the activity easier for non-experts. These are described as “democratised” roles.
According to the report, professionalised roles experienced twice the job growth and 42% faster wage growth than democratised roles since 2021.
That distinction matters.
AI does not push every wage in the same direction.
It can increase the value of people who use technology to expand expertise that remains difficult to replace.
And it can reduce the market value of people who mainly offer execution that has become widely accessible.
Automating part of the work changes the price of the bundle
A profession contains several tasks.
Some are rare and complex.
Others are repetitive.
Some require judgment.
Others follow clear patterns.
A salary compensates for the complete bundle.
When AI automates part of that bundle, a company may reassess the value of the entire position.
Imagine a professional whose work contains ten activities.
Four can now be performed almost entirely by a tool.
The other six still require a person.
The profession has not disappeared.
But the organization may conclude that the role requires less time, less specialization, or fewer workers.
That conclusion may be correct or incorrect.
The remaining activities may be more difficult and risky.
The professional may need to review automated output, handle exceptions, and take responsibility for a much larger volume of work.
Even so, the company may use automation as an argument for reducing the perceived value of the role.
Human work has not been eliminated.
It has been repositioned as a complement to the tool.
Complements often have less power when the organization treats the technology as the central element.
Productivity does not guarantee higher pay
AI promises higher productivity.
In theory, a more productive professional creates more value and can earn more.
In practice, the gains are not distributed automatically.
They may be used to raise wages.
They may also become profit, lower prices, expanded operations, smaller teams, or higher targets.
Who controls the technology and who possesses bargaining power influence the distribution.
A study by International Monetary Fund researchers published in July 2026 estimated, using observed usage data, that time currently saved by AI could be equivalent to approximately US$2.7 trillion in annual labour costs, or 3.4% of the combined GDP of the 86 countries analyzed.
The authors emphasize that this is an indicative measure of the labour cost corresponding to the time saved, not money automatically converted into economic growth. In developing economies, the value measured through AI use remained concentrated in a small group of professional occupations.
This demonstrates the scale of the opportunity.
It also demonstrates the scale of the dispute.
When AI saves an hour of someone’s work, who receives the value of that hour?
The professional?
The company?
The customer?
The provider of the technology?
There is no technological answer.
There is an economic negotiation.
A professional may produce more while earning the same
One of the least visible forms of wage pressure occurs when productivity rises but compensation remains unchanged.
Previously, a team completed ten tasks each week.
With AI, it completes fifteen.
The company can interpret this as additional capability.
Or it can immediately convert it into a new minimum target.
After several months, fifteen tasks no longer look like progress.
They become the expectation.
The professional works with more powerful tools, manages more deliveries, reviews more code, serves more customers, and assumes greater responsibility.
Their salary remains the same.
In this situation, technology has not nominally reduced compensation.
It has reduced compensation relative to the value produced.
This effect can be difficult to notice because the professional may feel that they are progressing.
They deliver more.
Receive praise for their efficiency.
Learn new tools.
But they do not participate proportionally in the economic gain.
Productivity without distribution may simply mean work intensification.
An NBER working paper associated greater AI exposure with longer working hours and less leisure, primarily in settings where AI complemented rather than replaced human labour. As an observational study, it does not prove that AI will lengthen hours in every context, but it warns that productivity tools can extend work instead of reducing it.
Smaller teams may negotiate more poorly
When teams shrink, we might expect the remaining professionals to become more valuable.
Sometimes they do.
In areas with genuine skills shortages, professionals capable of operating critical systems, integrating AI, and accepting complex responsibility can increase their bargaining power.
But smaller teams can also mean fewer alternatives for workers.
The company concentrates knowledge among fewer people.
It becomes more dependent on them.
At the same time, it expands responsibilities and creates a culture in which everyone fears the next reduction.
In that environment, professionals may accept more work without demanding proportional compensation.
Insecurity reduces the willingness to negotiate.
The market does not need mass unemployment to discipline wages.
It only needs professionals to believe that external opportunities are limited and that many candidates are available.
The AI wage premium exists — but not for everyone
Current evidence also points in the opposite direction from universal wage compression.
PwC found an average wage premium of 62% in 2026 for roles requiring specific AI skills. Those postings grew by 69%, compared with 9% growth across the broader job market analyzed.
The premium varied significantly by sector, reaching 118% in some industries and 16% in government and public-sector work.
This demonstrates that AI knowledge can create a significant economic advantage.
But it does not mean that anyone who uses an AI tool will earn 62% more.
The premium relates to scarce and explicitly required capabilities.
It may also reflect experience, industry, location, and job complexity.
A high premium also attracts new professionals.
Supply grows.
Tools become simpler.
Knowledge that was once rare becomes common.
Part of the premium may decline over time.
Learning AI remains important.
But building an entire career around the most popular skill of the moment may replace one vulnerability with another.
The market may divide one profession into two levels
As AI assumes more predictable tasks, a single profession may split.
On one side are professionals who operate tools, perform structured work, and review simple outputs.
On the other are professionals who define problems, integrate systems, evaluate risks, make decisions, and remain accountable for consequences.
The titles may be similar.
The pay will not be.
This division is visible in PwC’s distinction between professionalised and democratised roles.
When AI amplifies the specialist, judgment and contextual knowledge become more valuable.
When it enables non-specialists to perform an activity, competition increases and wage growth tends to be weaker. is not only losing a job.
It is remaining on the side of the profession that is becoming a commodity.
Experience may lose value in some tasks and gain it in others
For years, companies paid more for experienced professionals because they could perform tasks more quickly and accurately.
AI reduces part of that advantage.
A beginner can produce a reasonable first version.
Access explanations.
Explore a codebase.
Create tests.
Generate alternatives.
This narrows the gap in initial execution between experience levels.
But the technology does not automatically eliminate differences in judgment.
Experienced professionals understand consequences.
Notice inconsistencies.
Identify risks that are difficult to describe.
Understand the history of the system and organization.
Experience loses value when it consisted mainly of operational shortcuts.
It gains value when it represents perspective, decision-making, and accountability.
This helps explain why certain experienced roles are becoming more valuable while parts of intermediate work face pressure.
AI does not necessarily reduce the experience premium.
It changes what counts as valuable experience.
Junior work may face pressure from two directions
Early-career professionals may face two difficulties simultaneously.
The first is fewer entry-level opportunities.
The second is a lower price for the tasks they can initially offer.
When AI performs structured activities, companies may hire fewer juniors.
When they do hire, they may argue that the tools allow the professional to produce more from the first day.
Requirements rise.
Compensation does not necessarily follow.
A report by IMF researchers published in January 2026 found higher wages in jobs requesting new skills, particularly those related to IT and AI. But it also linked the spread of these skills to lower employment in occupations with high exposure and low complementarity with AI, creating particular risks for young workers.
This can create a paradox.
The market pays very well for advanced AI abilities.
But it puts pressure on the value of entry-level tasks that AI makes easier.
A beginner must invest more to enter a role with potentially weaker bargaining power.
Pressure may appear in the contract
Formal salary is not the only measure of the economic quality of a job.
A company may maintain monthly compensation while changing the employment relationship.
It may replace stable employment with temporary work.
Direct employment with independent contracting.
Internal teams with outsourcing.
Defined working hours with delivery-based arrangements.
Fixed compensation with variable payments.
Professionals may receive a similar gross amount while assuming more costs, risks, and periods without income.
AI makes it easier to fragment certain activities.
When work can be divided into smaller tasks, it becomes easier to distribute it among suppliers, platforms, or independent professionals.
The profession continues to exist.
Its economic protection declines.
Looking only at average wages can therefore hide changes in stability, benefits, and predictability.
In its June 2026 review of emerging evidence, the ILO emphasized that the major risks currently involve inequality, reduced opportunities for younger workers, autonomy, and job quality—not only the total quantity of employment.
Technology can weaken perceived authorship
There is another effect on bargaining power.
When a company believes that most of the result was produced by a tool, it may attribute less value to the person.
The professional completed the task.
But the organization thinks:
“AI did almost everything.”
That perception may be unfair.
The professional may have defined the problem, provided context, identified errors, integrated the result, and accepted complete responsibility.
Those activities are less visible than the final code or document.
When human value moves toward invisible work, professionals need to make that contribution understandable.
Not to claim authorship over every line.
But to demonstrate that the tool alone would not have produced a reliable result.
Someone who cannot explain their contribution risks appearing to be merely the operator of a system the company believes could be assigned to anyone.
Value is moving from execution to control
When execution becomes cheap, the market begins valuing other parts of the process.
Defining the correct problem.
Establishing criteria.
Selecting data.
Integrating tools.
Validating results.
Managing risk.
Making decisions.
Communicating consequences.
Taking responsibility.
These capabilities can protect bargaining power because they are more difficult to standardize.
But no protection is permanent.
Some of them will also be partially automated.
The strategy should not be to search for a task AI will never perform.
Such a prediction is impossible.
A more resilient strategy is to keep moving toward positions in which the professional understands the complete system, learns through change, and can assume new problems.
What protects a professional’s pay?
No skill offers absolute protection.
But several factors increase bargaining power.
The first is genuine scarcity.
Using a popular tool is not enough. Professionals need a combination of knowledge that is difficult to find.
The second is proximity to outcomes.
People who can demonstrate impact on revenue, costs, risk, quality, or customers tend to negotiate better than those assessed only by task volume.
The third is accountability.
The greater the consequences of a decision, the greater the need for people capable of taking responsibility.
The fourth is contextual knowledge.
Understanding specific systems, customers, regulations, and processes creates value that is difficult to replace immediately.
The fifth is mobility.
Being able to work across sectors, functions, or markets creates alternatives and strengthens negotiation.
The sixth is learning ability.
When one task loses value, the professional needs to build value in another part of the process.
What should companies measure?
Companies adopting AI need to avoid superficial metrics.
Lines of code.
Number of tasks.
Document volume.
Average response time.
Those indicators may increase while quality, understanding, and sustainability decline.
They may also encourage organizations to conclude that people have become less valuable simply because they produce more with tools.
More responsible measurement should consider:
value created;
quality;
rework;
incidents;
review time;
risk avoided;
learning;
customer satisfaction;
team health;
distribution of productivity gains.
When a company measures only production, it tends to convert every increase in capacity into a new obligation.
When it measures value, it can distinguish genuine productivity from work intensification.
Wages will not face equal pressure
There is no single wage future for the AI era.
Some professionals will earn less.
Others will earn more.
Some roles will lose prestige.
Others will become strategic.
Part of the difference will depend on AI capability.
Part will depend on work organization.
Part will depend on laws, unions, contracts, and institutions.
Part will depend on the bargaining power of each group.
An ILO analysis of aggregate productivity warned that without policies involving skills, digital infrastructure, social protection, competition, and collective bargaining, AI may widen productivity and income gaps among companies, workers, and countries.
Technology creates the potential gain.
Institutions help determine who receives it.
So, will the impact reach wages first?
For many professionals, probably.
Not necessarily through a direct pay cut.
It may arrive through weaker raises.
Higher targets.
Fewer people on the team.
More responsibilities.
Worse contracts.
Greater competition.
Fewer promotion opportunities.
Or declining value in the tasks that once supported a particular salary level.
For others, the effect will be the opposite.
AI will expand their capability, increase their responsibility, and make their combination of skills more scarce.
Those professionals may capture a significant wage premium.
The difference lies in how the technology relates to their knowledge.
Does it primarily replace what they offer?
Or amplify what they know how to do?
Does it make their work accessible to more people?
Or enable them to solve problems few people can assume?
The job may survive while the career loses value
This is the risk that unemployment headlines fail to show.
Someone may continue working.
Remain occupied.
Keep delivering.
And gradually discover that compensation is no longer keeping pace with productivity, responsibility, or the cost of living.
The job has not disappeared.
Its ability to sustain a career has weakened.
Preparing for AI therefore means more than avoiding dismissal.
It means protecting the ability to negotiate.
Making created value visible.
Developing complementary capabilities.
Moving closer to decisions.
Building professional alternatives.
Understanding who captures productivity gains.
Artificial intelligence can dramatically increase the value created by one person.
That does not mean it will automatically increase the value paid to that person.
Between productivity and wages, there is a negotiation.
In labour markets, technology has never been the only factor determining who wins it.