What happens to the career ladder when AI removes its lower rungs
The conversation about AI and jobs often seems sharply divided. Some people predict widespread job losses, while others say AI will create new opportunities and boost productivity. Both views have some truth, but neither tells the whole story.
A bigger question is how AI will change the way people gain the experience they need for the jobs that remain.
AI might not replace experts, but it could take away the work that helps people become experts.
Organizations now use AI for research, drafting, analysis, programming, customer interactions and administrative tasks. In the past, people early in their careers did much of this work. If these tasks disappear, productivity might rise at first, but a long-term skills gap could emerge.
Jobs Change Before They Disappear
AI will cause some job losses, especially in roles that are repetitive, rule-based, language-heavy or easy to digitize. But jobs include many tasks, so automating some parts usually changes the role instead of removing it. People still need to apply judgement, handle exceptions, manage relationships and make decisions.
That is why we should be careful with job forecasts. A company might not cut many jobs because of AI, but it could still change its workforce by hiring fewer people, allowing departures to go unfilled or raising expectations for those who stay. These changes might appear first in fewer new hires and slower promotions, long before they are visible in overall employment numbers.
Radiology is a good example of these complexities. In 2016, Geoffrey Hinton predicted that deep learning would outperform radiologists within five years. But by 2025, the Mayo Clinic had about 400 radiologists, which is 55% more than in 2016, and was using over 250 AI models. The 2026 US residency Match also offered a record 1,241 diagnostic radiology positions.
This does not mean automation cannot affect radiologists. Strong performance on one task does not mean an entire profession can be replaced. I see the same pattern in consulting. AI can speed up research and create a good first draft, but consultants still need to check assumptions, understand clients and apply judgement.
The Disappearing Bottom Rung
Early-career employees have traditionally gained experience by doing research, writing first drafts, analyzing simple data, recording meetings, listening to customers and handling basic problems. AI can now perform many of these tasks remarkably well.
Early signs suggest this is affecting hiring. The Stanford Digital Economy Lab found that employment among workers aged 22 to 25 in highly AI-exposed US occupations was 19% below where it would have been if it had kept pace with less-exposed occupations. Most of the gap came from reduced hiring, not increased separations. The researchers describe this as an early signal, not proof that AI caused the entire difference.
Canadian evidence is more cautious. Statistics Canada found no broad AI-driven employment decline through December 2025. However, most growth in coding-intensive work was among workers aged 30 to 49, while employment among coding professionals under 30 stagnated. Other economic and demographic forces may also explain the pattern.
Even with these uncertainties, workforce development remains a real concern. If new employees do not get the work that teaches them how a business operates, how will they gain the judgement needed for senior roles in five or ten years?
Productivity Is Not the Same as Learning
AI can also change how much expertise a role requires. If a system recommends the next action, produces the response, summarizes the history and identifies the likely answer, a person may become proficient at using the system without developing the underlying skills.
A 2025 Microsoft Research survey of 319 knowledge workers found that greater confidence in AI was associated with less self-reported critical-thinking effort. That is an association, not proof of skill loss. Medicine offers a stronger warning. In a 2025 observational study, experienced endoscopists’ adenoma detection rate in procedures without AI fell from 28.4% to 22.4% after AI-assisted detection was introduced. The study does not prove that AI caused the decline, but it suggests routine assistance can erode a skill when the tool is removed.
This does not mean people should avoid AI. Leaders need to decide which skills must remain and ensure that employees can spot weak answers, solve problems without the tool and recover when the system is unavailable. Organizations must distinguish between making expertise easier to access and allowing it to disappear.
Productivity Can Also Create Demand
Automation does not always reduce employment because lower costs can boost demand. James Bessen’s account of ATMs is a useful example. Between 1988 and 2004, the average number of tellers per urban branch fell from about 20 to 13. Lower operating costs also allowed banks to open more locations, and the number of urban branches increased by about 43%.
The point isn’t that higher productivity always creates jobs. Automating a task can lower costs, boost demand, and change a role before it removes it. Leaders decide the outcome by choosing whether to use saved capacity to cut costs or to improve and expand services.
When AI Changes the Organization, Not Just the Job
HubSpot’s October 2026 workforce reduction shows why the connection between AI and employment is not always direct. The company eliminated nearly 660 positions, about 7% of its workforce. CEO Yamini Rangan explicitly said the decision was not driven by AI-related efficiencies.
At the same time, HubSpot said it had shifted its strategy towards delivering customer outcomes with AI and reorganized with fewer management layers and clearer ownership. This does not mean AI replaced 660 people. It shows how AI can influence employment through strategy and organizational change, even when automation is not the stated reason for layoffs.
What CCaaS Tells Us About Contact Centre Jobs
I already see this shift in my consulting work with customer experience and contact centres. Contact Centre as a Service, or CCaaS, platforms can summarize conversations, suggest responses, automate workflows and handle routine questions with virtual agents.
In April 2026, Gartner reported that 63% of service leaders were gradually reducing staff through attrition, while 31% had implemented or planned AI-driven frontline layoffs. At the same time, 85% were adding responsibilities to frontline roles and 75% were moving agents into new roles. As routine contacts are automated, human agents handle more complex, emotional or unusual cases.
AI can also shorten the learning curve. Brynjolfsson, Li and Raymond studied 5,179 customer-support agents and found that an AI assistant increased productivity by 14% overall and by 34% for novice and lower-skilled agents. Agents with two months of tenure and AI performed as well as agents with more than six months of tenure without it.
CCaaS can reduce the need for entry-level agents while helping new hires become productive faster. The outcome depends on the design. AI that explains and coaches can help people learn faster, while AI that simply completes the task can remove that learning path.
The Question Is Bigger Than Job Loss
AI’s impact on jobs will not be the same everywhere. Some jobs will disappear, others will grow, and many will change. One major shift may be difficult to see in employment statistics: fewer entry-level roles and fewer traditional ways to build expertise.
Saving time today matters, but so does keeping the skills your organization will need in five years. Every AI project should ask two questions: What work can be automated, and what learning comes from that work?
How will we develop the experts we’ll need in the future if AI takes over the work people used to learn from? Are you building AI to replace entry-level staff, or to help people move up faster? How is your organization thinking about hiring and mentoring juniors in the age of AI? Share your thoughts in the comments.