I came across a term recently that should guide how we use AI.
The “chain of understanding”.
If you watch cop shows, you’ll have heard of the chain of custody – the process that ensures evidence can’t be tampered with.
We need something similar for AI.
AI can generate huge amounts of content – but our ability to absorb and verify it hasn’t changed.
So do we really read and understand what it produces? Or do we trust that it’s right?
Yesterday, I asked two different AIs the same question. They gave two confident but contradictory answers.
That’s the risk.
In many contexts, choosing the wrong answer has an impact radius – affecting millions in investment and rippling through supply chains.
The issue isn’t speed.
It’s whether you understand the logic underpinning a decision or how a program actually works.
That’s the chain of understanding.
AI can generate answers. It can’t take responsibility for them.
If you can’t explain it, you probably shouldn’t act on it.
Handcrafted insight 