Abstract:

In an influential 2006 article, Jeannette Wing advocated for the expansion of computational thinking. As a result of this advocacy, programs to encourage computational thinking have received considerable public support, with many being adopted around the world and in a variety of educational contexts. At the same time, concern over algorithmic bias has risen to public consciousness as disturbing examples of racial, gender, and other forms of discrimination have been discovered and then discussed in the media, raising grave concerns about the fairness of algorithms used in the criminal justice system, health care, government, education, and commerce. While the cause of algorithmic bias is not always known, data sets that reflect historical biases, data points that serve as proxy for identity markers, the inappropriate use of data sets, and feedback loops often play a role. The rise of machine learning, artificial intelligence, and the internet of things is likely to expand the possibility for--as well as the impact of--algorithmic bias. Without careful safeguards, computational thinking programs can, unfortunately, become a vector for entrenching and extending algorithmic bias. A careful interrogation of the principles of computational thinking will show that computational thinking can be taught without encouraging the proliferation of algorithmic bias, and recommendations for avoiding the problem will be suggested.

Keywords:

Computational thinking, algorithmic bias, data analysis, algorithms