This fascinating article shows pretty conclusively that at least some
people are optimizers, although maybe they're the exception. BTW, if you
ever need to have an A+ grade removed from your academic record, read the
article because it shows you how.

On a different note, I have some comments on why many people seem to be
rule followers rather than optimizers. Consider evolution as an AI
designer. Any AI designer faces two major problems:

1. Design a decision algorithm that improves utility, taking into account
the costs of computation.

2. Prevent misinterpretation and random drift of the utility function.

Problem 1 obviously implies using simpler subroutines when stakes are low, 
and more complex resource-intensive subroutines when stakes are high. The 
traditional justification for modeling people as perfect optimizers is 
that the model will match reality when stakes are high enough, and who 
cares about the low stakes situations?

But if you think about problem 2, you'll realize that there may be
a net advantage to following rules blindly even when the stakes are high. 

A perfect optimizer who behaves according to decision theory (or some
bounded-rationality version of it) is very vulnerable to small changes in
its utility function definition or the module responsible for interpreting
the meaning of terms in the utility function definition. Such a change,
say a bit flip caused by cosmic radiation, or the introduction of a new
philosophical idea, could cause the agent to behave completely counter to
the designer's intentions.

In the rule-based agent, on the other hand, the utility function 
definition and its interpretation are effectively dispersed throughout the 
set of rules. If the rules are designed with appropriate redundancy, it 
should be much less likely for a catastrophic change in behavior to occur.

Reply via email to