Comments by "Crazy Eyes" (@CrizzyEyes) on "Google Engineer on His Sentient AI Claim" video.
-
@janesmy6267 It is by definition, not how humans think. In psychology, it's called heuristic thinking. Humans use shortcuts to arrive at conclusions based on experience. They can notice patterns in things that relate to something similar they've done before, even if they're doing something else for the first time, and arrive at the correct answer with no prior experience. An AI, as we know it today, is kind of like training a dog (although honestly I believe that comparison does a disservice to dogs, but it's a difficult concept to explain). They must be explicitly trained or instructed how to deal with situations on a case by case basis. We call this algorithmic thinking. As of today, we get over this limitation (well, a little) by simply throwing billions of cases at supercomputers so they can crunch numbers and train themselves for so many cases or situations. But it isn't real heuristic thinking.
To put it another way, if there were a human student and an AI who both showed up to take a final exam: The AI would have "learned" and passed the test by simply memorizing all the answers, but the human (assuming they are a good student) would have learned by building blocks of fundamental understanding so they can arrive at the same correct answer without memorization.
3
-
1
-
1
-
1