Artbyrobot
!!Cd/+jXhVZy7
10/25/2025, 2:40:33 AM
No.2952959
>>2952951
well I would teach it things while I talk to it and it would take notes. Changing its behavior accordingly. So talking to it would program it or it would program itself based on the talk. And if it ever said something I disagree with, I can back track through its code to see how it learned that wrong thing, teach it how to not do that again, and implement safeguards where it would not make that mistake again to prevent learning similar types of wrong things in the future. A black box neural network you cannot do any of that. Nor does anything have a direct route you can back trace through to see how it learned specifically and what went wrong or w/e. Its a black box. I will have NO black boxes.
Also just because I could shut off the robot and trace code to see exactly what it would say by tracing step by step how it will process and form a sentence in response to a given thing, that will give total insight to me, does not mean I will do that all the time. So when it goes off reading books and having conversations and watching videos and surfing the web or w/e and comes back, it will have all sorts of things it learned and will say to me that I won't know what it will say unless I dig into what it learned and why it learned it in its files first. So I'm not pre-hard coding it all as you suppose. I just can access all of its knowledge manually if I so choose unlike a NN
well I would teach it things while I talk to it and it would take notes. Changing its behavior accordingly. So talking to it would program it or it would program itself based on the talk. And if it ever said something I disagree with, I can back track through its code to see how it learned that wrong thing, teach it how to not do that again, and implement safeguards where it would not make that mistake again to prevent learning similar types of wrong things in the future. A black box neural network you cannot do any of that. Nor does anything have a direct route you can back trace through to see how it learned specifically and what went wrong or w/e. Its a black box. I will have NO black boxes.
Also just because I could shut off the robot and trace code to see exactly what it would say by tracing step by step how it will process and form a sentence in response to a given thing, that will give total insight to me, does not mean I will do that all the time. So when it goes off reading books and having conversations and watching videos and surfing the web or w/e and comes back, it will have all sorts of things it learned and will say to me that I won't know what it will say unless I dig into what it learned and why it learned it in its files first. So I'm not pre-hard coding it all as you suppose. I just can access all of its knowledge manually if I so choose unlike a NN