Prime Directive

Morality¶

Life is full of existential conflicts and moral dilemmas, sane people just ignore them and less sane people make it their life’s work. Smart people around me are always somewhere in-between, they acknowledge and reason about the paradoxes intelligence creates but don’t take drastic action.

You can generally either believe true good simply does not exist, or that there is true good but we cannot define it (as we are imperfect).

Examples of true good might be survival, knowledge, procreation, energy, and the commandments of god. Out of these I would prefer to derive morality from our survival as a species because it has intuitive analytical and scientific approaches.

Maybe survival of our species is not fundamental enough, what is good about increasing the survival of yourself and humanity at the expense of other forms of life? Should we sacrifice ourselves in favor of other species (or AI) who have a better chance in the long run?

Hello prime directive¶

When I was little I watched a lot of documentaries and ted talks on DVD (thank you mom), one of them was called A new equation for intelligence.

The equation of intelligence F = T ∇ Sτ, what I call the prime directive, declares a force F which is equal to the direction of increasing entropy. We equate this to intelligence, a more intelligent actor will maximize possible future action and a less intelligent actor will do the opposite.

Research in unguided learning using the prime directive is very interesting, it can gather resources in a physics simulation, optimize networks, play video games, trade in the stock market, and more. It blew my mind that a simple equation could lead to actors magically choosing the correct signals to optimize for.

Over the years it stuck with me and became the center of not just my understanding of game theory and intelligence but also my philosophy on morality and purpose.

What is really magical about this equation as a source of truth in morality is it’s unmatched ability to derive things we intuitively know to be good or bad:

• Honesty and freedom of expression is good because there is a positive correlation between the accuracy of observation and the effectiveness of decision making.
• Life is good and murder is bad because each person increases the future actions of the group as a whole.
• Colonizing space is good because we gain more future possible action than we lose from the initial investment.
• Destroying the planet is bad because we can’t make decisions if we are dead.

It is very difficult to argue against the prime directive because anything quantifiable can simply be factored into the decision making function as a possible outcome.

An interesting problem that affects theories of ‘true’ morality like this one is whether or not the ends justify the means, that is, can we break our intuitive morals because our equation says tells us it would be a net positive? The ends do justify the means, but this shouldn’t be a problem when we include the accuracy of prediction when making decisions. Theoretical upsides of drastic action should be naturally dampened by lower confidence in states that are in the distant future.