Bartholomew Gallacher
Well-known member
- Joined
- Sep 26, 2018
- Messages
- 6,814
- SL Rez
- 2002
The general problem with AI when using neural networks is that we humans really don't understand much what's going on in there, and how it determines its results. An AI is basically an answer to a given problem where we do fail to create an algorithm, so instead we do use this approach. And aside it's nice name such a thing is really, really dumb because it can only do one thing well, and has no awareness about what it is doing. It's at best the intelligence level of less an insect, deciding stuff on an instinct, nothing more. Aside that neural networks are a simplified theory on how real brains do work, so much stuff has been cut out as well you can find in biological brains.
Such an AI is for humans always a black box, we are literally still in the beginning of understanding how an AI decides stuff at all. It's good if it works the way we want it to, but if not, then "debugging" it becomes a nightmare.
Such an AI is for humans always a black box, we are literally still in the beginning of understanding how an AI decides stuff at all. It's good if it works the way we want it to, but if not, then "debugging" it becomes a nightmare.














