This news item where the AI refuses a task because it is unethical or unfair as per it (or, to be technical, its programming) is interesting. That means AI-infused products, which will be more or less everything in the near future, will likely come with a built-in moral compass that will reflect their owners’ politics. And many a time, this bias may not be immediately or openly apparent.
The other thing they may have baked in might be opinions about how a historical event is to be interpreted, which art, writing, or poetry is to be considered good, what the standard of beauty should be, and other such subjective issues of preference and social conditioning, which may be passed on from the makers/owners to the system.
The system could therefore refuse to either obey the command given to it or choose to interpret the command in the context of the ethical system it has been based on, and thus structure its answer (if there is one at all) around that bias without properly and fully disclosing it a priori.
I am not sure that is a good thing?
Because I can see this polarising businesses and dividing the consumers along political lines. It has already happened along these in the kind of source from where one consumes the news. But this can soon start permeating every walk of life. I am not sure this is beneficial to humanity in the long run.
Perhaps we need some universal laws like Asimov’s Laws of Robotics to ensure fairness and even-handedness in our everyday products and services, at least as far as the fundamentals go. What do you think?
