The GPS problem: how AI is quietly eroding human capability

The analogy that should worry every leader

Ask someone under 30 to navigate from one side of a city to another without their phone and watch what happens. It's not that they're less intelligent than previous generations — it's that GPS has meant they never needed to develop the spatial reasoning, landmark awareness, and mental mapping that older generations built through years of navigating without assistance.

This is the GPS problem: a tool so useful that it doesn't just replace a task — it prevents the development of the underlying capability. And it's exactly what AI is doing, right now, across almost every domain of knowledge work.

Where capability erosion is already happening

This isn't a future risk — it's happening now, in organisations that have moved quickly to integrate AI into their workflows:

Domain	AI tool	Capability at risk
Writing	AI drafting assistants	Ability to structure arguments, articulate ideas, find the right words under pressure
Data analysis	AI analytics tools	Pattern recognition intuition, ability to spot anomalies, statistical reasoning
Software development	AI coding assistants	Deep understanding of how code works, ability to debug without AI, architectural judgment
Decision-making	AI recommendation engines	Judgment developed through wrestling with hard decisions, risk intuition
Research	AI summarisation tools	Ability to evaluate sources, synthesise conflicting information, form original views
Customer service	AI support agents	Empathy, de-escalation, handling genuinely novel situations

Why this is a business continuity problem, not just an HR one

Most organisations frame skill erosion as a talent development concern — something for HR and L&D to worry about. But it is also a fundamental business continuity risk.

Consider: if your AI-assisted document processing system goes down, and the team that previously handled it manually has forgotten how to do so — or been replaced by a smaller team that never learned — you don't just have a technology outage. You have an operational capability gap that cannot be resolved quickly.

The military understood this long before AI. They continue to train soldiers in map-reading despite universal GPS availability. They continue to train pilots in manual flight despite sophisticated autopilot systems. The reasoning is simple: when the technology fails, the human needs to be capable of taking over.

What smart organisations are doing about it

The answer isn't to slow AI adoption — the competitive cost of that is too high. The answer is to adopt AI in a way that augments rather than replaces human capability. Organisations getting this right are doing several things:

• Rotating new staff through manual workflows before introducing AI assistance, so foundational skills are built first
• Running quarterly "manual mode" drills — treating it like a fire drill, normalising the idea that AI won't always be available
• Requiring human explanation of AI outputs — if a team member cannot explain why an AI recommendation makes sense, they're not allowed to act on it
• Preserving documentation of manual processes rather than retiring them when AI takes over
• Hiring for domain expertise alongside AI proficiency, not instead of it
• Measuring human capability independently of AI-assisted performance metrics

The deeper question for leaders

Every leader adopting AI should be asking two questions, not one. Not just "how do we get the most from AI?" but also "what human capabilities are we at risk of losing, and how do we protect them?"

The organisations that answer both questions will have the best of both worlds: AI-accelerated productivity and the human capability to function, adapt, and make sound judgements when AI isn't available or isn't right.

The organisations that only ask the first question are building a dependency that, sooner or later, will cost them.