Conflict between Machine Learning and Societal Values


Profiled article

Moses, L. (2018). Is Your Algorithm Dangerous? – IEEE Technology and Society. [online] IEEE Technology and Society. Available at: [Accessed 15 Nov. 2018].

About the article author

Lyria Bennett Moses is an Associate Professor and Director of the Allens Hub for Technology, Law and Innovation at UNSW Law.


In this week’s article profile, we will be discussing a piece by Lyria Moses from IEEE’s Technology and Society magazine. The article explains the author’s thoughts on how to determine if a given machine learning algorithm is appropriate. Ultimately, the author notes that the particulars of whether an algorithm is appropriate is highly context dependent. But, she also points out two widely relevant concerns: is the algorithm answering the correct question and is it displacing any social values? These two concerns raise important questions meriting further discussion. For example, if an algorithm provides a statistically significant but seemingly unrelated correlation (e.g. if height is found to positively correlate with academic success), should it still be used? With regards to the second point, at what point should social values supersede statistical significance?

Related Links

A brief and broad overview of machine learning:

An article on predictive policing, which utilizes machine learning to predict crime.