Algorithms work in ways that are often unknown, even to their own developers. #artificialintelligence #machinelearning #ethics

Algorithms are diverse and have many different functions. In computer science terms, an algorithm is an abstract, formalized description of a computational procedure. 

Algorithms vary in their analytic characteristics, such as how their mean-time or best-time performance varies with the size of the data sets over which they operate. 

Algorithms have a name. Some algorithms are so pervasively used that they have names –either those of their inventors (Dijkstra’s algorithm, the Viterbi algorithm, Gouraud shading, or Rivest-Shamir-Adelman) or conventional names (e.g. QuickSort, Fast Fourier Transform, Soundex, or sort-merge join).

Algorithms in Context

Meanwhile, the public discussion of algorithms usually emerges at the intersection. 

Algorithms are often part of debates around data volume and big data, and the ways in which data is used to feed algorithms that derive patterns, that are then used to make decisions at many different levels (corporations, institutions, states, etc.). 

Others focus on velocity, highlighting how algorithmic automation allows for high-speed processes like high-frequency programmed trading in stock markets, where computer systems take action without human intervention. 

Algorithms are also prominent in discussions on how technology impacts the labor market, turning humans into resources deployed according to programmed responses. That’s the case in ride-sharing services like Uber.

Algorithms are also often used to obscure debates and displace responsibility. For instance, the 2016 U.S. presidential election result was partially attributed to two ‘algorithmic’ failures, for instance. ‘First, Facebook’s emerging as an influential news provider and its algorithm’s failure to filter out ‘fake news’. Second, the failure of most election forecast institutions to predict the result of the U.S. election even when using state-of-the-art algorithms performing aggregation, big-data analysis and forecasting.’ (Acman et al 2017)

The Social Impact of Algorithms

At the moment, much of the debate about algorithms focuses on the context of big data or machine learning techniques, but not the social, ethical and legal impact of algorithms.

A better approach to frame algorithms is to view them as socio-technical systems, which recognizes both human and nonhuman components that act together in such a way that are recognized as part of a single system. 

Why? Algorithms work in ways that are often unknown, even to their own developers. They produce what some have called ‘unknowns’ – an analysis of data that are known and understood in some terms (i.e. the data set’s patterns and regularities) but unknowable in others (i.e. the domain that the data represents). 

For instance, when a credit card company deems a particular purchase or stream of purchases ‘suspicious’ and puts a security hold on a card, the company cannot explain exactly what was suspicious – they know that there’s something odd but they don’t know what it is. 

Algorithms shift and evolve in deployment and often remain hidden in the name of trade secrets. To understand Google’s algorithm is frankly out of the question as implementation and strategy continuously evolve.

In Houston, algorithms were used to evaluate teachers’ performance and Houston teachers were able to overturn the system on due process grounds. They successfully argued that because the vendor considered the evaluation system a trade secret, they were denied the right to use the data to understand or improve their performance.

Therefore, the limits of an algorithm are determined by a combination of social engagements as much as technological  constraints.

Below is an initial list of areas of algorithmic social impact, pleading for explainability, fairness, accountability, regulation and a better understanding of the role technology plays in society, away from technophilia and technophobia, is already urgent.