← All insights ← →

Is AI Eroding Skills and Creating Technical Dependency?

Published: 5 September 2025

Is AI Eroding Skills and Creating Technical Dependency?

The Opportunities and Risks of AI in the Enterprise

AI is far more than a technology buzzword by now. It is changing decision processes, value chains, and business models at a pace that surprises many.

The benefits look obvious: higher efficiency, faster analysis, automation of routine work, and the ability to make complex relationships visible.

This euphoria carries a real danger, though: AI can weaken human agency. Two developments deserve particularly close attention.

  1. Skill loss («de-skilling»): capabilities that are no longer actively trained get lost.
  2. Vendor dependency («lock-in»): companies drift into technical and commercial dependencies that are hard to escape.

A third dimension is moving into focus: the systematic cutback of junior and training programmes. Many companies hire fewer entry-level staff or scrap whole programmes, with a major impact on the future skills base.

When People Unlearn What Machines Take Over

Automation relieves the burden; that is a genuine benefit. It speeds up processes, cuts cost, and reduces error rates, at least at first glance.

It also changes how people use their own capabilities. «Use it or lose it» is a principle well known from sport, medicine, and aviation.

  • In medicine, studies show that doctors who regularly rely on AI-assisted diagnostic systems perform significantly worse in tests without AI support. Their ability to spot anomalies on their own weakens.
  • In aviation, we see similar effects. Pilots who rely heavily on autopilot lose manual and cognitive sharpness. In critical situations, when technology fails, this can have fatal consequences.

For companies this means that as more processes get automated, the long-term ability to question systems critically or intervene manually in an emergency declines.

Junior Programmes in Retreat: A Quiet Crisis

The impact on the talent pipeline is even more serious. Many companies are cutting back junior programmes or hiring far fewer entry-level staff.

  • A Stanford study finds that entry-level positions in fields such as software development and customer service fell by 13 per cent over three years, precisely in the areas where AI is used most intensively.
  • SignalFire documents that large tech companies hired 25 per cent fewer entry-level staff in 2024 than the year before.
  • The World Economic Forum Future of Jobs Report 2025 confirms the trend: 40 per cent of employers plan to reduce junior roles as tasks shift to AI.
  • According to LeadDev, more than half of engineering leaders expect fewer junior hires in the long run, leaving a gap in practical experience.
  • The Wall Street Journal already reports on companies that have scrapped entire junior programmes, believing AI can fully replace simple tasks.

The consequence is a quiet crisis in talent development. Without young talent, companies will lack the specialists to fill middle management roles or drive new business models within five to ten years. Organisations risk not only skill gaps but also a marked loss of innovative capacity.

Cognitive Dependency: When People Trust Technology Blindly

Beyond the factual loss of skills, there is a psychological dimension too: automation bias. People tend to trust machine output more than their own judgement, even when they have doubts (see my article: Does AI Influence Our Decisions?). In practice, this means:

  • Flawed analysis gets accepted without review.
  • Control functions weaken because «the AI must know better».
  • Simple false assumptions compound across entire decision chains.

This creates a new form of cognitive dependency for companies. It is less visible than technical lock-in, but just as dangerous: it undermines the ability to decide independently and critically. Most of us know a version of this effect from our own lives: mental arithmetic, reading a map, remembering phone numbers, and similar skills that fade with disuse.

Regulatory Framework: Oversight Is a Duty, Not an Option

Regulators have recognised the risk too.

  • The EU AI Act requires companies to guarantee effective human oversight for high-risk applications.
  • The NIST AI Risk Management Framework in the US stresses the need for governance, risk assessment, and human control.

For boards and executive teams, this means human oversight must be an integral part of every AI strategy. It is not just a regulatory requirement; it is above all a strategic safeguard.

Vendor Dependency: Lock-in as a Strategic Weakness

The second major risk is less psychological and far more practical: technical lock-in.

Many AI systems are proprietary. Data sits in closed environments, interfaces are not standardised, and switching providers is expensive and complex. The consequences can be severe.

  • Companies lose flexibility over time.
  • Pricing and innovation cycles get dictated by a handful of vendors.
  • Strategic decisions become dependent on third parties.

In dynamic markets, this can be dangerous over time. Companies that cannot switch lose speed and competitiveness.

Strategic Countermeasures

Boards and executive teams need to identify these risks early.

Protecting skills

  • Simulations and training keep capabilities from fading.
  • Job rotation and peer learning broaden knowledge and build resilience.
  • A culture of questioning: staff need the mandate and encouragement to scrutinise AI output critically.

Keeping junior programmes alive

  • Continue junior programmes, even where short-term demand for simple tasks has dropped.
  • Expand mentoring and onboarding systematically.
  • Build a long-term talent strategy aimed at future leaders, not just current efficiency.

Avoiding lock-in

  • Use multi-model strategies to stay able to switch between vendors.
  • Secure data sovereignty: company knowledge belongs in its own infrastructure, not in proprietary systems.
  • Negotiate contractual exit options to stay able to act if needed.

Recommendations for Action

Boards should weigh the following steps.

  1. Set strategic guardrails: define where automation makes sense and where human oversight remains essential, even if it means forgoing short-term cost savings.
  2. Scrutinise investments: when assessing AI projects, look beyond efficiency gains to portability and talent development.
  3. Fund talent development: allocate budget and resources to junior programmes, training, and mentoring.
  4. Demand risk reporting: request regular updates on technical dependencies and skill risks, on the same footing as cyber or financial risk.

Conclusion: Balance Decides Future Readiness

Artificial intelligence is a powerful tool. It raises efficiency, cuts cost, and opens new business lines. Companies that chase short-term gains alone put their long-term strategic agency at risk.

  • Without skill retention, the company becomes vulnerable to errors and failures.
  • Without junior programmes, the next generation of specialists and leaders is missing.
  • Without technical independence, the company loses flexibility and negotiating power.

The future belongs to organisations that keep this balance. They use AI without devaluing people. They invest in talent, even when the temptation to automate is strong in the short term. And they build their systems so they can switch providers and decide for themselves at any time.

This is not a purely technical task. It is a strategic leadership decision.

References