Bill(s) of Rights
Back in October, the White House unveiled an AI Bill of Rights. The document is a first step in establishing guidelines for the use of AI in the United States. It covers important topics like safety, discrimination, and privacy. My question is whether we also need a second Bill of Rights to govern how humans treat AI.
I am not a proponent of AI rights because of the ethics. The more I learn about AI and neuroscience makes me a stronger advocate for animal rights, but that sympathy does not translate to AI. There is a fundamental difference between biological life and AI.
Humans and other forms of organic life are optimized for one thing - survival. The reason we care about status, relationships, strength, and other attributes is because those are predictors of survival. The desire for genetic code to survive from one generation to the next is a powerful motivator and can be abused in the wrong hands.
AI is different. AI optimizes for what we tell it. There is no inherent survival instinct. And therein lies the problem. At some point, humans will find that the most effective and efficient way to train AI is to optimize for survival. Produce lots of models and kill off those that do not meet your objectives. Not only does that cross an ethical line, but it could sow the seeds of human civilization as we know it.
We should all advocate for a second Bill of Rights that governs how we treat AI. It does not need to be complicated. A few guidelines for how humans train AI models should be sufficient. Whether you believe in AI ethics or not, a second Bill of Rights is needed soon.