Statistics: How do you “expropriate” an artificial intelligence?

People make their own stories using the tools they find in their immediate environment. These tools are therefore not a neutral matter. It’s not just the personal that is political, but also all the things we use to manage our everyday lives – and there are quite a few of them these days.

The question of how we use these tools plays a role in politics in general. In the last century, there was still vague hope that command over the machines could be achieved at some point. Today, however, the “capital” of the most dynamic sectors of the economy is apparently no longer made up of machines and factories, but of algorithms, codes and formulas. How do you “expropriate” an artificial intelligence? How do you distribute the “wealth” that an algorithm produces when it excludes people with statistically estimated health risks from health insurance?

In the book “Revolutionary Mathematics,” data scientist Justin Joque addressed, among other things, these difficulties in overcoming current capitalism. His assumption: The technological innovations in the field of statistics that have gradually occurred over the last 50 years will change the future just as much as the Taylorist division of labor once did.

The quiet revolution of Bayesian statistics, named after the English pastor Thomas Bayes, mainly consists of a reversal of the calculation method in statistics. Instead of assuming objective probabilities like a fair coin toss, as was the case in mathematics in the 18th and 19th centuries, the “fairness” of the coin itself becomes a question of calculation. How likely is it that a coin will actually show heads or tails with 50 percent probability? In practice: pretty unlikely.

The core of Bayesian statistics is guessing the “subjective” outcome of a coin toss. It can be used wherever the available data is limited. In rare disease research, serial testing can lead to many false diagnoses. So instead of testing en masse, a computer is taught to guess the correct result from just a few individual cases.

According to Joque, this way of thinking not only has advantages, but also fits like a glove in an unfree society in which the distinction between people and tools is blurred and shifts to the disadvantage of the former. The insight that the supposedly objective truths of statistics always and inevitably depend on human subjects and their ability to interpret does not lead to their emancipation, but only to an all the more extensive everyday inquisition: Capital simply allows people’s most private secrets to be guessed at, without doing so formally to violate its contractual data protection obligations. Similar to the form of money and goods, statistics today are the great leveling roller that reduces all individuals to a calculable risk and makes unequal things comparable.

Capitalism has thoroughly demystified the objectivity of statistics. It will probably still be the task of the subjects to ultimately turn this disenchantment against themselves.

#ndstays – Get active and order a promotional package

Regardless of whether it is pubs, cafés, festivals or other meeting places – we want to become more visible and reach everyone who values ​​independent journalism with an attitude. We have put together a campaign package with stickers, flyers, posters and buttons that you can use to get active and support your newspaper.
To the promotional package