Regulator Warns Tech Companies of Big Data Bias

Recently the Federal Trade Commission has taken it upon itself to issue a stern warning to companies who are trying to take advantage of big data and this warning consists of being mindful of your biases. Increasingly in todays world big tech companies need to consider “whether their data sets represented the user population, whether their algorithms were insulated against social bias, and whether the predictions they generated were accurate.” When sorting through the massive sets of data it can be extremely difficult to detect biases against certain groups of individuals. This is but one of many problems that plague large data sets.

The article urges these tech companies to remember the golden rule of logic amongst IT companies and that is “Correlation does not equal causation”. Clearly managing increasingly large amounts of data is no simple task however being able to correct the mistake of an obvious bias is crucial to catering to the many needs of your consumers. The article goes on to say “Companies should remember that while big data is very good at detecting correlations, it does not explain which correlations are meaningful”. If a tech company is not paying attention to these certain biases than they are eliminating a large chunk of potential clients. These firms may not even be aware that they are committing any such bias which makes these problems so difficult to solve.

One example that the article identifies to give the reader more of a real world perspective is Google’s ability to predict flu trends. Google sought to use search queries on flu related topics to predict flu outbreaks ahead of government estimates. Google initially was accurate in their attempt, however, in the years that followed overshot the prevalence of the illness leading to the question whether or not data mining should be used to make large estimates. After reading the article I feel as though I am armed with enough knowledge to critique these types of trends such as the Google over prediction of flu outbreaks. I could understand how a company would want to use the given data that they have amassed to make meaningful predictions that could help many people. However, at what expense? Google extrapolated data to make too large of an estimate leading to their subsequent mistake in predicting flu trends. Even a massive company such as Google is susceptible to making these critical mistakes which should be used as a reminder to all tech companies that these biases are out there and that not detecting them can lead to major problems.

URL:http://blogs.wsj.com/digits/2016/01/07/regulator-warns-tech-companies-of-big-data-bias/?mod=ST1