The decisive ball of the last tennis match between. The nth out of play that is because of any rival of Barça de Hansi Flick. Or automatic postures correction to avoid injuries like the one that took Carolina Marín del Oro in the Paris Games. For all this, computer or computer vision is already used. This branch of (AI) is also behind the automatic identification of friends in photos, and other incipient revolutions, such as advanced robotics or the discovery of new proteins. But the coveted object of artificial vision (VA) is humans: according to a, the main scientific journal, the vast majority and studies (papersin jargon) and derived patents focus on people, to identify their different parts, what they do and the environments where they move.
Researchers from several universities in the United States and Europe have collected thousands of papers and also thousands of patents presented or registered since 1990, when research in artificial vision still limited with science fiction. Then extracting information from an image or video was a challenge for machines. It still is, as visuals demonstrates to confirm that we are human. However, it has been advancing both in the ability to analyze data and ability to interpret them. There are many fields in which. But it also has its reverse and is huge.
Supported by about thirty keywords, some as explicit as surveillance o facial recognition and others not as obvious, such as iris o airportthe researchers analyzed 19,000 papers Above it is presented at an important annual conference of the sector that has been held for more than 30 years. They did the same with 23,000 registered patents derived from those works. The former release the advances achieved, the latter protect them so that only their authors can take advantage of it.
The results are summarized by the authors in their article Nature: “We discovered that 90% of articles and 86% of derived patents extracted data related to human beings.” Most of the works, more than two thirds, were about obtaining information from the human body as a whole or of certain parts of it, especially the face. Another portion of articles and patents (18% and 16%, respectively) extracted data on human spaces such as work, home or traffic areas. A lower percentage, but nothing despicable, 1% (5% for patents) enters the category that researchers called “socially relevant human data” and that would seek that machines apprehend mental states, economic level, cultural belonging …
By putting these results in the timeline is when the emergence of an entire ecosystem of mass and automated surveillance technologies is detected. In the early years of the 1990s they have only found a few hundred investigations and patents derived in the field of the VA and only half were oriented to analyze human data. Two decades later, the annual production of papers and patents had tripled and already 78% focused on human data. Combining the absolute increase in research in the vision of machines and the relative of humans centered, the science of surveillance is quintupled.
“It is difficult to establish a clear cause of the significant increase in the production of articles and patents related to surveillance,” acknowledges Trinity College in Dublin (Ireland) and senior author of this study, in an email. However, it points out some factors: “The technical advance of the field of artificial vision, the lack of a critical and socially conscious culture within the field, the institutional structures that reward surveillance or magical thinking around technology that presupposes that surveillance technologies are the solution to political and cultural problems.” It also mentions the great increase in financing for the interest invested by powerful organisms that benefit from surveillance, either by power, control or economic benefit.
Another of the data that have revealed the authors is the concentration of the investigation in the field of the VA. Ahead of all countries, the United States stands out, which produces more research than all others combined. A certain distance, but far ahead of the third classified (United Kingdom) is China. I shouldn’t miss. On the one hand, they are the two technological superpowers, regardless of the object or use of the technology in question. In addition, the authoritarian character of the Chinese regime and the connection between the American military and defense industry with the most pointed science would explain the rest. In fact, two of the most relevant actors of the last generation AI (not necessarily related to artificial vision), Palantir and Anduril ,. The first already develops with immigrants in its objective. And Openai, the creator of Chatgpt, also also signed the United States.
In a comment also published in Naturethe researcher at Monash University (Australia), Jathan Sadowski, who has not participated in this study, stops a result of it. The authors analyzed the evolution of the language used in the papers and patents, detecting a tendency to hide its purpose with increasingly neutral words or eliminating them without more. For example, the reference to humans is increasingly frequently appearing even if they are the subject of studies and are replaced by terms such as objects.
“Referring to humans as objects could be considered as a simple abbreviation for engineering,” writes Sadowski. It would fit into a more general tendency of technological language. But for this scientist, these dehumanization practices cannot be ignored lightly: “These technologies are created in a political and economic panorama in which the interests of large corporations and military and police institutions have an enormous influence on the design and use of AI systems.”
