12, May 2026

In 2026 THECIS celebrates its 25th anniversary. To celebrate this, we are starting a 25th Anniversary Blog series where we ask prominent individuals to write a blog to provide perspective on a topic related to innovation and entrepreneurship. This Blog is by Stephen Murgatroyd. Stephen teaches at the University of Alberta and is President and CEO at futureTHINK Press.

 Blog 107:  AI’s Impact on Entry Level Work Demands Innovation

Artificial intelligence is not simply changing work; it is rewiring the traditional bridge between learning and employment. For decades, entry-level jobs performed a hidden but essential economic function. They converted classroom learning into practical competence, introduced new workers to professional norms, gave employers a low-risk way to assess talent, and helped individuals form occupational identity. As we argue in recent major reports from the LearningCITY Collective’s Productivity Project, that bridge is now weakening. What appears to be a labour-market issue is also a learning-system crisis. It affects a wide range of jobs, including professional work (e.g. accounting).

The disruption is structural. AI systems (especially agentic AI) increasingly absorb the very tasks that once populated entry-level roles: drafting routine documents, processing transactions, scheduling, coding simple scripts, handling first-line customer service, summarizing information, and basic analytics. These were not glamorous tasks, but they were developmental tasks. They were where people learned judgment, workflow discipline, communication under pressure, and how organizations actually function. If machines now do more of this work, the number of “learn by doing” opportunities contracts.

This creates a paradox. Employers say they need adaptable, digitally capable talent. Yet the mechanisms for producing such talent are shrinking. We are seeing the rise of the experience trap: candidates need experience to get hired but cannot get it because starter roles have been automated, outsourced, or redesigned into fewer, higher-skill positions. Many job advertisements seek a specific qualification and three years of experience. Emerging employer evidence suggests AI may be constricting entry-level hiring pipelines. In a 2026 survey, 21% of firms reported already freezing entry-level recruitment and 47% expected to do so by 2027.

The first implication is for schools, colleges, and universities. Many learning systems still operate on a twentieth-century model: teach theory first, then assume the labour market will complete development. That assumption is increasingly unsafe. Institutions can no longer be content with knowledge transmission alone. They need to become capability accelerators. They also need to assess for capability and skill, which means rethinking learning and assessment in fundamental ways.

This means redesigning programs around applied performance. Every credential should ask: what can graduates do, demonstrate, solve, build, communicate, or improve on day one? Work-integrated learning, simulations, client projects, apprenticeships, portfolio assessment, challenge-based learning, and interdisciplinary problem solving should move from the margins to the core. If real entry-level jobs are scarcer, learning environments must reproduce some of the developmental value those jobs once provided.

The second implication is for employers. Too many organizations still treat training as a discretionary cost rather than a strategic asset. Canada spends less than almost all OECD countries on employee development. In an AI economy, this is backwards. Firms that reduce junior hiring without replacing developmental pathways may gain short-term efficiency while hollowing out future leadership pipelines while at the same time reducing their adaptive capacity.

Forward-looking employers will build “AI plus human” entry routes. Instead of expecting graduates to arrive fully formed, they will create rotational programs, mentored project roles, internal academies, digital apprenticeships, and structured progression pathways. Junior employees should learn how to supervise AI outputs, verify quality, interpret ambiguity, build client trust, and integrate tools into workflows. These are high-value human capabilities that complement automation rather than compete with it.

The third implication is for public policy. Much current policy remains trapped in what might be termed “innovation theatre”: announcing AI hubs, pilot projects, glossy strategies, and isolated grants while ignoring the deeper operating system of talent development. Real innovation is not another conference about the future of work. It is redesigning how citizens continuously gain recognized capabilities across a fifty-year career. That requires ecosystem thinking and reimagining how we assess and credential learning.

An effective response would connect five elements:

  1. Skills intelligence infrastructure. Real-time labour-market data linked to learning opportunities and career pathways.
  2. Recognition systems. Micro-credentials, competency assessments, and trusted ways to validate skills acquired through work, study, or self-learning. Moving to a competency passport for learner, assessment-on-demand using an agreed competency standard, a reinvention of prior learning assessment.
  3. Lifelong learning finance. Portable training accounts, tax incentives, and co-funded reskilling supports for workers and employers.
  4. Regional talent partnerships. Colleges, employers, unions, municipalities, and community groups co-designing pathways into growth sectors. Rather than having 100 different certifications in a field like marketing (currently the case in Alberta), there would be a standard certification with progressive layers of competency (rather like the practice in nursing – The National Licensure Nursing Examination).
  5. Inclusion mechanisms. Ensuring rural communities, newcomers, mid-career workers, and those displaced by automation are not left behind.

This is where many jurisdictions fail. They treat education policy, labour policy, innovation policy, and economic development policy as separate silos. But AI disruption cuts across all four. The answer cannot come from one Ministry, one college, or one employer acting alone.

There is also an important cultural shift required. We must stop fetishizing credentials and start valuing demonstrated capability. A degree will remain valuable, but it cannot be the sole passport to opportunity. In a faster-moving economy, people need multiple re-entry points: chances to retrain, prove competence, and move again. It is worth noting that the reskilling | upskilling sector in Canada is currently a $2.9 billion sector and is anticipated to be a $10 – $14 billion sector by 2030.

The opportunity, if we choose it, is significant. AI can reduce routine work, personalize training, map skill gaps, accelerate onboarding, and make continuous learning cheaper and more accessible. Used well, it could democratize development rather than concentrate opportunity. But this only happens if we intentionally design systems around human growth and capabilities.

The real risk is not that AI destroys all entry-level jobs. The deeper risk is that it quietly removes the developmental ladder into adulthood, contribution, and economic security. When that happens, productivity stalls, inequality widens, and social trust erodes.

The challenge before us is clear: move beyond innovation theatre and build learning ecosystems that match the realities of an AI economy. The question is no longer whether AI will change work. It is whether our learning systems, employers, and governments can change fast enough to keep people, organizations and our learning systems moving forward.