Social science for algorithmic societies

Machine learning algorithms pervade modern life. They shape decisions about who gets a mortgage, who gets a job, and who gets bail, and have become so enmeshed in our political and economic processes that some scientists argue we are witnessing the emergence of “algorithmically infused societies.”

In a new perspective piece for Nature, SFI External Professor Tina Eliassi-Rad and her co-authors ask how social scientists can investigate algorithmically infused societies, which may require very different methodologies than social sciences have traditionally deployed.

“The existing toolkit of social theories and measurement models was not created with the deep societal reach of algorithms in mind, and may thus not apply to human societies that are permeated by algorithms,” they write.

They call attention to major challenges for measuring social phenomena in an algorithmically infused society, and outline five best practices computational social scientists can follow to mitigate “the harmful consequences of (mis)measurement.”