People familiar with personalities like Elon Musk and his AI venture would have probably come across certain promises like "building the truth AI (A.K.A unbiased AI)" through xAI, but I find all of it comical because that is literally one thing AI can never be.
AI can be a lot of things but being unbiased is not one of them.
AI cannot be unbiased because people responsible for it's development are incapable of being unbiased.
This is a reality that calls into question what definitively represents "truth" and in most cases, people acknowledge that the concept of a universal truth is flawed because what's closest to the truth in any scenario is completely up to personal perspective, understanding and effectively judgement.
This means that there's no way to attain a 100% consensus on what's true, it's not a factor of objectivity.
It's rather all subject to individual interpretation.
This reality opens everything up for bias, regardless of how hard we try to eliminate it.
Sometimes bias are intentionally embedded and other times, it's completely naturally occuring.
I think this is something we ought to acknowledge early on because it impacts how we approach and treat AI systems.
We have to be prepared for bias in AI systems. Some systems will be more favorable to certain concepts, beliefs, and even people, and that should not fundamentally be a problem.
That said, we have to be prepared with implementation of safeguards and regulations of far these bias are permitted to exist, specifically what the greater impact should be.
Because when left unchecked, it grows to cause great damage and that's something we cannot have.
Posted Using INLEO
