Will AI developers be on the bench of future Nuremberg-like trials?
The Nazis basically said ‘i was just following orders’ - i.e. ‘chain of authority defense’. Will ai devs follow the same pattern of defense once/if an ai swarm does something serious that harms a lot of people beyond a reasonable degree? ‘I was just coding a reward function’, ‘i was just perfecting an algorithm’, ‘i was just trying to get the work done so i could get home’ - i’m curious if new laws will have to be enacted and retroactively applied to ai devs, similarly to what happened with the Nuremberg trials
Comments
One of the cornerstones of jurisprudence, going back to Ancient Rome if not earlier, is that you cannot retroactively apply laws. It's expressly prohibited in most jurisdictions; new laws come into force only after they're signed into law.
Besides, most AI devs work for corporations, and their work belongs to the corporation, so the corporation itself would be liable. At worst you might see the fabled "Corporate Death Penalty" applied, but I doubt it.
Maybe i’m recalling the youtube videos I saw on Nuremberg incorrectly, but i recall something on the lines of: the concept of ‘crimes against humanity’ being applied retroactively
When you start finding yourself asking these types of questions, it is a very strong sign to slow down and step back.
Whether you listen to your intuition or go full speed ahead... Well, that's on you. I know what let's me sleep at night.