The "inherent political nature" of machine learning algorithms

Around 1200 BC, the Chinese Shang Dynasty had already manufactured thousands of bronze wares for daily and sacrificial purposes. In this example of early large-scale production, the bronze foundry process required intricate planning and coordination among a large number of workers who completed separate tasks in a precise sequence.

A thousand years later, similar complex craftsmanship was also used to make the famous Terracotta Warriors and Horses. These statues were "manufactured by an assembly line production system that lays a foundation for large production and commerce."

Some scholars have speculated that the form of these prescripTIve-work technologies played a very important role in the formation of Chinese society. Together with other factors, they allow people to accept the social philosophy of the bureaucratic system that emphasizes the hierarchy, and also make people believe that there are simple and correct methods for everything.

When industrial chemical plants were born in Europe in the nineteenth century, even the strong critics of Engel's capitalism admitted that, regardless of whether the economic system is capitalism or socialism, large-scale production is a necessary condition for centralization. In the 20th century, theorists such as Langdon Winner extended this idea to technology. He believes that (for example) the atomic bomb should be regarded as an "inherent political product" because its "dead property requires it to be controlled by a centralized chain of rigid command levels."

Today, we can extend this idea even further. Consider machine learning algorithms—the most important general purpose technology used today.

The "inherent political nature" of machine learning algorithms

A key feature of machine learning algorithms is that their performance improves as the data increases. Therefore, the use of these algorithms forms a technical force to process information about people as recordable, recallable data. Like large production systems, they are "inherently politic" because their core functions require certain social behaviors and hinder other social behaviors. In particular, the spread of machine learning is directly opposed to the individual's desire for privacy.

A system based on public access to information about individual members of society seems to fit socialists such as Amitai Etzioni (communitarians) who believe that privacy restrictions are enforced by social norms. means. But unlike socialists, algorithms do not care about social norms. They only focus on making better predictions, and this can be achieved by turning more and more areas of human life into extensible data sets.

Algorithm evaluation is not new. Scholars such as Oscar H. Gandy warned that we are transforming into a recorded and ranked society and demanded more accountability to correct technically induced errors. However, unlike modern machine learning algorithms, old evaluation tools can be quite thorough. They make decisions based on relevant normative and empirical factors. For example, it is no secret that carrying many credit card debts damages one’s credibility.

In contrast, new machine learning techniques exploit deep data sets to find relationships between things that are predictable but not fully understood. In the workplace, algorithms can track employees' conversations, where they have lunch, how much time they spend on computers, on the phone, or in meetings. With this data, the algorithm can develop complex productivity models that far exceed our common-sense intuition. In an algorithmic elite system, what the model requires and what becomes an excellent standard.

Despite this, technology is not fatal. We decide technology before technology decides us. Business leaders and decision makers can develop and deploy technologies based on their organizational needs. We have the ability to put a privacy net around sensitive areas of human life, protecting people from the effects of harmful data usage, and asking the algorithm to balance the accuracy of predictions with the values ​​of fairness, accountability, and transparency.

But before we follow the logical flow of natural algorithms, more elitism is unavoidable. This change will have a profound impact on our democratic system and political structure. If current commercial and consumer cultures continue, we will soon have more similarities to virtuous politics and socialist traditions than to our own individualism and liberal democratic traditions. If we want to change trends, we must place our own political responsibilities before technology.

65W Wall-Mounted Power Adapter

65W Wall-Mounted Power Adapter,12V5A Wall-Mounted Power Adapter,Ac-Dc Power Adapter,Switching Power Adapter

Guangdong Mingxin Power Technologies Co.,Ltd. , https://www.mxpowersupply.com