043 - Cognitive Trade-offs
I just received confirmation yesterday that another of my research papers will be published later this month in an issue from the Polish Academy of Sciences (PAN). It addresses an often overlooked factor about Cognitive Bias, the fundamental trade-off governing the degree to which it must be applied. The more complex the problem, the more cognitive bias is required for humans.
The human brain doesn't scale, it must fit within the human skull when addressing any and all problems an individual may encounter throughout their lives. Over 200 documented cognitive biases offer the mechanism by which this can be achieved, as each bias reduces the complexity of a problem in distinct ways, and those biases are easily and most often unconsciously combined.
However, these cognitive biases have functional limits, bounded by factors of evolutionary history and the far slower rate of adaptation found in long-lived biological systems. Beyond groups the size of a historic tribe the utility they offer begins to decline, as illustrated in research such as Robin Dunbar's contribution, now known as "Dunbar's Number".
This trade-off becomes a critical factor to consider when two things are true:
-
The size of a group is very large (Governments and large companies).
-
A scalable system with human-like thought and learning is available (Norn, being rebuilt for scalability now).
Humans are built to automatically apply heavy cognitive bias to hyper-complexity, as well as time scarcity since the two are functionally identical in resources demanded over time. Humans can build systems that scale in the ways necessary to overcome that trade-off, and that is a big part of the work we do at AGI Laboratory.