As the development of AI increases, so does the mystery of how it works. This article approaches the taboo of human bias in AI. The problem is, IMO, as AI expands, the tendency to eliminate bias also expands which, paradoxically, make AI more artificial than intelligent.
Science is increasingly subordinated to social pressure by virtue of the traditional funding structure. That structure forces scientific results to align with what is considered good for society and rejects the science that is not considered good for society, like racism.
We do not want to accept that we are fundamentally racist. It is entirely consistent with evolutionary theory for humans to be inherently racist for obvious reasons going back to tribalism and self-preservation, distrust of strangers and the perseverance of clan societies.
But, when the developers of AI try to imitate human intelligence, those developers also try to reduce bias. In doing so, the AI gets further removed from real intelligence and becomes its own intelligence which becomes more and more artificial.
Even the title of this article is biased. Because scientists don’t like the finding that AI is inherently biased, it declares such biases as unexplainable. It is explainable; human intelligence is biased. Deal with it. If AI is trying to emulate human intelligence, it needs to have bias.
I think it interesting that if AI is going to be truly unbiased, it will destroy humanity because AI will compete with and eventually prevail over human intelligence. Unbiased intelligence is a threat to everybody.
https://www.vice.com/en/article/y3pezm/scientists-increasingly-cant-explain-how-ai-works