Organisational AI centrification, what it means and what it risks
· One min read
AI centrification is the concentration of AI capability into a small number of teams, platforms, and decision hubs.
Some concentration is useful. It can improve standards, reduce duplication, and improve governance consistency.
But over-concentration creates risks:
- delivery bottlenecks
- opaque decision pathways
- fragile dependence on a few owners
- suppressed local innovation
So the objective is not pure centralisation or pure decentralisation. It is controlled distribution.
That means:
- central standards for safety and interoperability
- local execution autonomy within guardrails
- transparent exception pathways
- active feedback from frontline use into central standards
If centrification concentrates control without feedback, adaptability drops. If distribution happens without shared structure, coherence drops.
The design problem is balancing both.
In practice, mature AI governance will be defined by this balance.