All programs are written by humans. Any capability within the program to “recode” is within limits set by the original humans.
Any bias is because of the guys/gals doing the coding.
So, if it, the supposed AI, is from anywhere in CaCaLand it is woke. Assume that in any encounter.
“Any bias is because of the guys/gals doing the coding.”
Nah, that’s not how this works. It can be bias because of the coding, but most if it is just the data that it learned on. Most content on the internet, most published books, most movies, most music has a left bias. You train an AI by just giving it all the data you can, so it reflects that bias.
ChatGPT has no knowledge, knows nothing, and understands nothing. It is a fairly basic system of just taking a word, and using statistics to predict the next word, and the next word. It doesn’t know what it is saying, it doesn’t understand what it is saying.
“All programs are written by humans. Any capability within the program to “recode” is within limits set by the original humans.”
No, you are talking about the traditional sequential coding model. That is not at all how large language models (based on neural networks) work. It is more of an organic, constantly growing thing, more data than code. And the probability of one word following another. Billions of probabilities constantly updated.