First you have to justify the hubris in the assumption that scientists are a necessary condition for intelligence.
People who think AI will be solved by gradient descent need to explain how scientists make discoveries using nothing else.
-
-
-
It’s the opposite. Intelligence is a necessary condition for scientist, so show that scientist work only using gradients. Not the other way around.
- Näytä vastaukset
Uusi keskustelu -
-
-
Rule based logic is ultimately implemented as gradients inside our brains.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
To soar in the sky, with wings that can't move, To heat up your bones, sans blood and sans food, To talk to a friend, you can't hear or see, But think not like us—no way that could be.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
you've brought up the subject I've been trying to push: develop first natural intelligence
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
the common factor in Science is making things unnecessarily complex and that is the result of lack of intelligence
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
Discoveries are a gradual process. Newton didn't discover gravity the very first moment. It was a hypothesis which he finally solved gradually, through trial and errors and updating his beliefs and internal errors. So did Einstein.
-
Ah, OK, so they did it by gradient descent.
- Näytä vastaukset
Uusi keskustelu -
-
-
Assumption: whatever "intelligence" is is a smooth dynamical system. Every smooth dynamical system over N variables can be seen as at most N independent gradient descents. QED. Can we find these losses is a separate issue :)
-
Both statements are false, and the conclusion doesn't follow.
- Näytä vastaukset
Uusi keskustelu -
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.