I’m confident that bias can be identified and eliminated. This is not easy and will need layers of ML beyond task based algorithms. The current direction is forms of Deep Learning.
This means we have a massive process to identify all the forms of human bias. But once that reaches a functioning level it can be applied with incresing context sensitivity. People just can’t do this. Even someone actively working to consciously stop bias from entering all statements will have bad days when bias reappears. In America, as an example, at our best we are a deeply racist culture. For most people this is not conscious and that just makes worse.
The whole process of being “woke” is realizing this. But, that has to happen every day and it requires structural change in our language. We have the segment of the population that cannot handle that at all. They simply can’t grasp that reality and are, as a result open to manipulation because of their confusion and the anger that results from the perception of being accused.
In short I don’t think we have any way to move forward except with augmentation via AI. That will be our next evolutionary step.