@JoeJ Thank you for your feedback!

What?

@taby If you get 50% always, it is a bad result. It is not how NNs work.

I’m trying to say that not all networks are created equally. The standard FFBP neural network does just fine on the XOR problem, and image classification of known images, but not the classification of unknown images.

taby said:
I’m trying to say that not all networks are created equally.

Correct. This is the current state of our tech.

Theoretically you can have a big enough fully interconnected network and it can handle just anything.
Take the brain as example. Various billions of people on this planet have the very same brain as a structure. Very same starting network. But there are no two brains to be found that repeat the same synapses.

You need a very very big computer though. But this is due to the limitations of our current technology. In the future our tech could improve enough. Every big country promises a quantum computer “very soon”. Nature proven that a supercomputer that can do just anything imaginable can fit inside a human head, regarding size.

Have you seen Ghost In The Shell? They connect wires to real live neurons. Researches already work on this too.

I hope it is a serious research, not just a marketing trick to make money on markets - https://corticallabs.com/

The research in AI takes many different paths. MY fav speculation is the Ghost In The Shell version of transhumanism. Though i would keep my brain clean of implants for the longest i can. I just would love to see this revolution happening. But i would not put wires in my brain until the very last moment.

If only one single of these paths succeeds, the practice of programming by hand will die. And the replacement of humans will begin.

Augmentation.

Terminator vs. Lawnmower Man sucked up by Borg Cube 0x 7of9, grown from microbot covid vacceine registering program terminated.

Sorry, I just got here. What is the task you're trying to solve that would take 10 years to program @NikiTo

@aganm This is not really relevant. Saying that it is too long is enough.

Don't optimize on CPU performance, optimize on your own development time.