Which Is Best

The problem with making the best choice in any situation is that it presupposes perfect knowledge of all future outcomes of the choice. So you should always make the choice that gives you more choices. Even if it hurts.

Because you're still learning to be you.

In Q learning, the technique used to create AlphaGo and many other formidable AI, the AI model learns how to play games at a superhuman level by breaking the task down into two parts:

• •Predicting the outcome $y$ at any point if I do action $a$
• •Predicting which outcome $y$ is the most desirable to achieve my actual goal (winning the game)

With these two elements, a goal, and extreme focus you can learn to do anything at a never before seen level over your lifetime. The question is what does it mean to win the game of life?

Can you ever really know what you want out of life? My hypothesis is that if you decide the answer is "no" then you will live a much more fulfilling life. Because you don't possess a time machine nor an infinite universe machine you can't have seen all the possible lives you could live and therefore can't even begin to make a decision "which is best" .

So the optimal strategy is to give yourself more exploration at every chance. Because you are still learning to be you, focusing on optimizing a local minima which has given you a large amount of pleasure for a long time may be stopping you from moving to a global minima that gives you 100x the payoff with 1/2 the effort.

You just don't know. 🙂