In the 2002 movie Minority Report, a complex algorithm and omnipresent means of observation combine to predict who will commit a crime, just in time to catch the perpetrator (not yet) red handed, thus preventing the felony from happening.
While the technology introduced in the movie remains (to the best of our knowledge) a dystopian spectacle of the future, big data analysis has been capable of astonishing inferences for several years now.
Given sufficient amounts of information, statistical correlations between different variables can be modelled. Without understanding their causal relation, unknown and more intimate data points, such as your gender, personality, sexual orientation, drug use, happiness, religious and political views can be inferred from more or less publicly available data points such as your Facebook likes, Instagram photos and the comments you leave on any number of different platforms.
The secret world of predictive psychometrics
One such project, YouAreWhatYouLike (by the Psychometrics Centre at the University of Cambridge), predicts your personality by recording your likes and referring to the personalities of other people who like similar pages to you.
Similarly, CrystalKnows uses “proprietary personality detection technology” and public data “to tell you how you can expect any given person to behave, how he or she wants to be spoken to, and perhaps more importantly, what you can expect your relationship to be like.”
In doing so, they are able to take the information we share online from various sources and contexts, and use it to predict underlying our personality patterns, without making the underlying algorithm transparent, and without asking for our consent.
How will it impact you?
Big data analytics can have real consequences for individuals, for instance when we’re applying for jobs, college admissions or credit. This should raise concerns if neither the methods of data collection nor the algorithms at the core of the analysis are transparent and obvious to us, the public.
That said, it is also true that big data techniques can yield convenient benefits such as improving and customising the services we use, and many applications may in fact have the potential to mitigate the consequences of discriminatory harm in each of these areas of life.
The danger is that left unchecked, such technological developments can also lead to the replication of discriminatory practices carried over from “the real world”, with unhelpful social norms becoming encoded into algorithms that perpetuate inequality and entrenched biases – and all beyond our knowledge or control.
Even the White House is weighing in
It’s this tension between the positive potential and possible harm arising from Big Data analytics, that was the central topic of the third report on Big Data: A Report on Algorithmic Systems, Opportunity, and Civil Rights, published by the White House at the beginning of the month.
In it, the authors propose to adhere to the principle of “equal opportunity by design”, that is to consciously design data systems that prevent discrimination and support equal opportunities.
Equal opportunity by design requires engineers of big data analysis systems to pay close attention to biases that could creep into the data analysis at various points in the process. From the initial point of data collection, to the algorithm used to infer personality traits, the paper highlights that both have the potential to be biased, “incomplete, incorrect or outdated” (White House, 2016).
By making the sources of the data, the algorithms and the conclusions transparent to those whom they pertain to, engineers can reduce the negative impact that their designs could potentially bring about.
If you’re interested in the topic of big data analysis and the privacy concerns that revolve around it, you can check out the following resources:
- Frederike Kaltheuner, a PhD candidate at the Amsterdam School of Cultural Analysis talks about the privacy harms of predictive data analysis in her eloquent and informative 2016 talk.
- If you are interested in the concerns, arguments and case studies considered in the White House report mentioned above, you can find it here.
- To learn more about how big data analysis is used in politics, have a look at this Guardian article on how Ted Cruz used Facebook data to gain an advantage over Trump in Iowa.