“You’re right… I should have… I’ll fix that.”
This sentence is harmless when it comes to code. It becomes chilling when it comes to the real world.
AI at the heart of our work
We use AI extensively in our work: software development, data analysis, automation, decision-making.
Yes, AI fixes code. Yes, it adjusts a DAX formula. Yes, it apologises… and tries again, better.
But let us ask a real question: what is “I’ll fix that” worth when the damage is already irreversible?
- A bug in production can be fixed.
- A wrong KPI can be recalculated.
- A missile, once launched, cannot be “debugged”.
When AI enters critical decisions
The news surrounding Anthropic and the Pentagon should push us out of our technological comfort zone. We are no longer talking about AI as a simple assistant, but about AI embedded in critical decision chains.
The real risk is not that AI makes mistakes. The real risk is giving it the right to make mistakes without sufficient human safeguards.
The questions every decision-maker must ask
Decision-makers, CIOs, developers, data analysts, executives:
- Who bears final responsibility?
- Where does AI autonomy end?
- At what point must humans take back control — before, not after?
AI excels at “fixing it afterwards”. Leadership is about preventing the irreparable beforehand.
And you: how far are you willing to delegate, when the cost of a mistake is no longer just technical, but human?
Every time AI tells you “You’re right… I should have… I’ll fix that”, remember: had you let it act alone, some mistakes would have been irreversible.