This article may highlight a growing trend in the credit scoring industry but its conclusions are absolutely ludicrous. The author draws comparisons between weighing and individual’s actions (i.e. spending history, payment frequency) by a computer and prejudices used to determine credit worthiness in the past by individuals. Isn’t this the entire problem the algorithmic method solves? The comparison is baseless on its face.
Additionally, if any factors used to determine a consumers credit worthiness are “biased,” “weblining” and “creating credit castes” than how exactly are lenders supposed to determine a consumers ability to repay credit? I’d like the author to think outside there very narrow view they’ve adopted and determine what they might consider when determining whether to lend money to someone they are not well acquainted with and see if the author can come up with a better and more effective solution for all stakeholders.
Excellent points made Rachel. I have seen far too many people with advanced degrees forget causation/ethics and be dazzled by shiny black box algorithms and data they dont understand. We need some rules around the usage of this data, and the new algorithms. AI/ML will only map the biased data you feed it.
Comments are closed.