For the past two years, AI has been portrayed as a massive job-replacement engine. Headlines predict entire professions becoming obsolete, and some industries are indeed changing rapidly.

But this narrative tells only half the story. An increasing number of studies and real-world deployments show a more complex reality: in many cases, adopting AI costs more than keeping human employees.


When economic intuition breaks down

Research by Peter Cappelli, a professor at Wharton, highlights a recurring pattern: what looks profitable in a slide deck often becomes far more expensive once deployed at scale.

Automation appears simple on paper. In real workflows, hidden costs quickly emerge.


The Ricoh case: a documented example

A well-known case analyzed with Harvard Business Review involves Ricoh’s attempt to automate insurance claim processing.

Key figures:

  • 6 people involved for 12 months (including 3 external consultants)
  • $500,000 in development and deployment costs
  • Initial finding: AI processing cost 3× more than humans
  • After optimization: ~$200,000 per month in AI expenses
  • Internal comparison: higher than previous payroll
  • Employment impact: 44 employees → 39
Hardly a case of massive workforce replacement.


Why AI can increase costs

According to Cappelli, AI boosts productivity but does not eliminate work in the short term. Instead, it creates new, often underestimated tasks:

  • human supervision,
  • exception handling,
  • error correction,
  • quality control,
  • system maintenance.
Work shifts — it doesn’t vanish.


Large-scale evidence supports the pattern

A report from MIT analyzing hundreds of AI projects found that:

  • 95% show no measurable financial impact
  • only ~5% generate clear value
  • success is concentrated in digitally mature organizations
AI delivers ROI only when deeply integrated.


What we see at Leadkong

At Leadkong, we observe that AI creates the most value when it augments teams instead of replacing them.

The most successful deployments:

  • focus on well-defined tasks,
  • keep humans in the loop,
  • measure total cost of ownership,
  • reallocate time toward higher-value work.


Conclusion

AI does not eliminate work overnight.
It reshapes it — and often introduces a costly transition phase.

The evidence is clear: AI transformation is primarily human and organizational, not just technological.