So aside from Wikipedia which is a publicly user maintained service which has become pretty reputable … the majority of the ‘facts’ that LLMs collect (about 75%) is all collected from privately controlled websites with curated content that is managed and maintained by corporations. And of all that content, most of it is also manipulated and controlled to make people either angry, mad, frightened, sad or anxious.
They’re teaching the next AI on our negative impulses, greatest fears and worst anxieties.
That would be a more honest representation of human culture rather than the curated content that is constantly manipulated and controlled by a private corporation.
So aside from Wikipedia which is a publicly user maintained service which has become pretty reputable … the majority of the ‘facts’ that LLMs collect (about 75%) is all collected from privately controlled websites with curated content that is managed and maintained by corporations. And of all that content, most of it is also manipulated and controlled to make people either angry, mad, frightened, sad or anxious.
They’re teaching the next AI on our negative impulses, greatest fears and worst anxieties.
What could go wrong?
Yes. Better if they collect it from personal blogs running on people’s PCs 👍
That would be a more honest representation of human culture rather than the curated content that is constantly manipulated and controlled by a private corporation.