The alarm you have to earn
An article on how we learn: facts are one memory, repetition is another, and enough repetition compiles a skill into an energy-efficient, instinctive mode called intuition. Outsource the repetitions to AI and the alarm never gets built. His own confession makes it concrete: his intuition whispered that 99.9% was too good to be true, and he wanted to believe.
// trace: where this idea came from
- ↳ video diary @ 2:43 (la repetición vuelve el saber instintivo)
- ↳ video diary @ 5:53 (la confesión: mi intuición me lo decía y yo quería creer)
- ↳ Entry 186-1: The whispered exam (la fuga que la intuición olió antes que el código)
The article he’s mid-reading splits learning in two: the factual kind, pure memorization, and the kind built by repetition, kin to motor memory ▸ 2:07. Repeat something enough and it migrates to “un modo más eficiente energéticamente”, runnable instinctively, which is to say it becomes intuition ▸ 2:43. The article’s warning follows directly: outsourcing memorization and computation to AI erodes capacities that are always worth having ▸ 3:04. The calculator was the first draft of the problem ▸ 3:26; Julia states the current draft: “dependiente de ChatGPT, tú ya no confías en tu intuición o mismo tu sabiduría” ▸ 3:43.
The test case is small and perfect: a calculator says 5 × 10 = 500, and something in a trained mind fires before any checking happens, that’s too high ▸ 5:27. That alarm is not free. It exists only in people who did the multiplications by hand enough times for wrongness to have a feel ▸ 5:38. Skip the repetitions and you don’t just lose speed, you lose the sensor.
sin repeticiones no se instala la alarma →
Then he supplies the diary’s own exhibit A, unprompted. During the eye-research experiments, the ones the leak inflated, his intuition was already talking: “estaba dando muy bien para ser real”, something is wrong here, “pero al mismo tiempo yo estaba como que quería creer” ▸ 6:04. The alarm fired; wanting overrode it. His conclusion is the article’s thesis lived forward: “siempre es importante trabajar una intuición” ▸ 6:12. He hasn’t finished the article; the RAM-like short-term memory of a first lecture, sharp for ninety minutes and gone by the next class ▸ 6:17, is where the bookmark sits…