166 - Decision-making Bias

A question I've been asked many times now is why the ability to detect and measure cognitive bias from both human and AI sources is important. One reason is the ability to cut through more of the "Noise" as Daniel Kahneman terms it, to isolate individual sources and composite patterns of bias as they are expressed over time and in sequence.

For those unfamiliar, "noise" refers to the inconsistency of human decision-making. This may be traced back to a complex tangle of interacting cognitive biases, provided that you have systems able to detect and measure individual cognitive biases as they are expressed in a given medium.

For example, this can be turned into a metaphor where you're the captain of a ship at sea. You have a destination you need to reach, but the weather and waves can add inconsistency to your trajectory if you lack the means of course correction. Early sailors used the stars to provide this correction, and we have more advanced means at our disposal now, but corporate board rooms and government committees more often than not have no meaningful methods of performing such course corrections.

Cognitive biases offer around 200+ different ways that decision-making in such board rooms, committees, and other governance structures may be thrown off course. If not detected and measured then these errors in judgments can only be read as "noise", inconsistencies absent any specific sources that may be well-isolated.

In the real world, when these kinds of errors in governance play out it often takes multiple years for the consequences of these errors to become highly visible, and the extremely late course corrections that follow tend to over-correct, in turn causing a different but equally predictable set of complications in the years after that. As this plays out over time many governance structures effectively give up, settling into poorly performing solutions thanks to repeated failures to adapt.

Much of history's scientific advances can be traced directly back to the ability to detect and measure something first being introduced. The case of detecting and measuring cognitive biases using systems that outperform the average human in a tiny fraction of the time is no different. It offers us new means of quickly and directly making those course adjustments in any decision-making process of consequence, reducing the influence of cognitive biases at scale.

Work on our own team's system which first demonstrated that capacity, comparing it directly to human performance on the same task, is still ongoing, but has continued to progress since last summer. The irony remains that most investors continue to make terrible decisions while showing no interest in addressing the very reason why that is.

Allow yourself to drift too far off course and you may as well list Wilson the Volleyball as your chief advisor.

Would you rather hire a taxi that only takes you to the correct city or one that takes you to the right address?