Like it's known that information is gotten from scraping sites on the internet, what if one of the web pages has wrong information and was also scrapped that will make the information wrong and what of a case of controversial topics of two opposing sides wouldn't that affect effectiveness in the long run of placing them in Humaniods.
The problem with that view is the model is trained, just like a human, on the information. Do you just learn from one input?
I don't, but I do have headache when there isn't similarities on the same aspect in different sources I researched, like in solving an accounting question that I didn't understand in school, so I decided to cram a method, seeing the other method made it a lot harder, so how will that be handled in your opinion ?