acm-header
Sign In

Communications of the ACM

ACM TechNews

How Do You Fight an Algorithm You Cannot See?


View as: Print Mobile App Share:
A black box symbolizes a lack of transparency in how algorithms come to conclusions.

Activists in Germany designed the OpenSchufa platform to determine how potential borrowers' credit scores are algorithmically generated by Germany's Schufa credit bureau.

Credit: cbronline.com

Activists in Germany have facilitated a method to assess the accountability of proprietary algorithms via a crowdsourcing platform.

They designed the OpenSchufa platform for the purpose of detailing how potential borrowers' credit scores are algorithmically generated by Germany's Schufa credit bureau.

Since the platform's rollout, several thousand people have donated their financial scores, and the activists determined the Schufa algorithm is significantly error-prone, creating relatively negative scores without any negative proof.

OpenSchufa is just one of a number of movements by theorists pushing for accountability of "black box" algorithms, to ensure their decisions can be explained.

Software engineers can test algorithms to determine if their output matches the expectations of a test set, but they currently lack insight into how the algorithms reach their final decisions.

From TechCrunch
View Full Article

 

Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account