A new research effort at the National Institute of Standards and Technology (NIST) aims to address a pervasive issue in our data-driven society: a lack of fairness that sometimes turns up in the answers we get from information retrieval software.

A measurably “fair search” would not always return the exact same list of answers to a repeated, identical query. Instead, the software would consider the relative relevance of the answers each time the search runs—thereby allowing different, potentially interesting answers to appear higher in the list at times.

Software of this type is everywhere, from popular search engines to less-known algorithms that help specialists comb through databases. This software usually incorporates forms of artificial intelligence that help it learn to make better decisions over time. But it bases these decisions on the data it receives, and if those data are biased in some way, the software will learn to make decisions that reflect that bias, too. These decisions can have real-world consequences—for instance, influencing which music artists a streaming service suggests, and whether you get recommended for a job interview.

 …