- Intro
- We know that human mind is limited, so we use shortcuts
- A lot of attention has been given to bias, and some has been given to noise. The difference is beyond the scope of this article
- Heuristics are grounded in some optimization, but not full. They are ‘good enough’
- Theit use can be seen as an example of why we need to offload to today’s AI, but even our current LLMs use heuristics
- They also face a capacity problem
- And we won’t farm all of our heuristics out to AI anytime soon
- We want to see an MLB player catch a fly ball, he is estimating the direction, speed and parabola of the ball. We hire this ability on the best calculation of heuristics, as an optimized fly ball catcher would arbitrage the enjoyment of baseball to 0 immediately
- Thus we need to understand heuristics to play our ape games. Forward.
- Anatomy of a heuristic
- Game theory > bounded rationality
- humans are only so rational in decision making - they make a decision that is good enough
- Necessity of heuristics
- Dangers of heuristics
- How to use heuristics