Is Forgetting the Secret to Mastery?
Why Forgetting Is Undervalued in AI
We tend to treat forgetting as a flaw in humans. In AI Machine Learning, it’s a core feature.
Intelligence — biological or artificial — isn’t about storing more. It’s about ruthlessly deleting what doesn’t matter.
“Pruning synapses. Regularisation. Knowledge distillation. Quantisation.”
All sophisticated ways of saying: stop memorising noise. In practice, this means:
reducing precision so the broader structure — the essence — becomes clear
removing information that doesn’t contribute to the signal
The same holds for us. The best experts aren’t the ones who memorise every rule in the book; they’re the ones who have stripped their mental model down to what actually matters.
I was reminded of this when I had to re-take the driving “Highway Code” for my motorbike licence. It was surprisingly difficult, even after twenty years of driving without a single accident. My brain had correctly discarded a hundred of rarely used rules and kept only what keeps you alive on the road.
Most experts do exactly that in their domain.
What we choose to keep matters. What we discard — deliberately or over time — defines the quality of our intelligence.
What do you think?
#AI #Intelligence #Memory
