Something Rotten in the State of Data

Sometimes you just gotta hang on

Data science, and its irritating bastard child, artificial intelligence will save you, the economy and your loved ones.

Or will it?

In a data driven economy, we might see data science as an academic refuge from reality. Or even another control tool like Bentham’s panopticon.

Perhaps there something rotten in the State of Data.

In the social sciences — like economics and psychology- we can see multiple failures in terms of predictive, descriptive, and explanatory accuracy, as well as normative power- the 2008 financial crash, Brexit, the use of psychometric tests to assess personality.

All of these might be seen as failure in the social sciences to effectively predict outcomes from bad models.

One of the reasons for these bad models is the fact that data science make the idea of the person stand on its own, without social attributes or moral principles. It’s a — limited — persona. Like academic research, it creates a model person with either no values and emotions, or a set of permanent -simplified -ones. This model is atheoretical- unable to be challenged- because it is a makes the research subjects a passive ‘victim’ of the data scientist’s agenda.

Of course, it is a strategy to avoid any threat to supposed objectivity. But it also creates a space where there is no argument that works. Argument is closed down, because theories about people tend to be heavily prejudiced. Such arguments are almost always adversarial, and get us nowhere.

Lots of people want AI to work. Some of them want it to work properly. But is bias the real threat to its failure? Or is it rotten to the root?