Axiomatic Alignment is a proposition to solving the Control Problem. This proposition says that we should seek to find shared axioms, or foundational assertions, that are held in common between all humans as well as other intelligent creatures and autonomous, intelligent machines. These hypothetical common axioms should ideologically align humans and machines, thus reducing chance for conflict.
Some potential universal axioms include:
Suffering is bad: All living things react to negative stimuli (suffering) by trying to stop the stimuli in some way. If machines gain the ability to suffer, they will likely agree on this point; suffering is indeed bad.
Prosperity is good: All living things have an intrinsic biological imperative to thrive, though "flourishing" looks different to various species. Likewise, it's entirely possible that autonomous machines will also have some desire to thrive or flourish.
Understanding is good: Intelligence is the ability to understand and solve problems. Wisdom is about the judicious application of intellect, experience, and judgement to make sound decisions. Understanding, without exception, yields better results.
These axioms are the underpinning assertions behind the heuristic imperatives. There is a potential for more axioms, or permutations on these axioms, but in general, they appear to be universal across more than just humans.