AI has named a problem leaders have faced for decades: Context Rot.
It happens when thinking quietly goes stale, while the world keeps moving forward.
Anthropic described Context Rot this week in their article on AI agents: “as the number of tokens in the context window increases, the model’s ability to accurately recall information from that context decreases.” In practice, that means memory drifts, fragments, and clings to outdated data - until decisions are made with yesterday’s picture of the world.
You’ve probably noticed it yourself in a long chat with AI. At some point it “loses the thread.” You say, “Remember when we talked about X?”, and it can’t tie back. That’s context rot in real time.
That phrase struck me, because humans face the same thing - when our thinking quietly goes stale while the world moves on.
Directors can easily fall into human context rot:
- Relying on strategies that made sense 3 years ago but don’t today.
- Using old mental models in a market that has shifted.
- Believing the environment is stable, when it’s actually been changing underneath us.
AI engineers fight context rot by pruning stale data, refreshing memory, and resetting context windows.
Leaders need to do the same.
Build your own “context hygiene”: revisit assumptions, rescan your environment, and prune ideas that no longer serve.
AI has a patch for context rot. Directors need a practice for it.
AIGovernance AILeadership BoardDirectors AIforBoards
