Artificial Intelligence (AI) carries many benefits, but also many risks. Therefore, eliminating human supervision can be a very bad idea. It is a fairly recurring topic that has been put back on the table after China announced that it is testing the use of AI to advance its satellite-based surveillance systems.
Many experts have expressed concern about the possibility of these systems being used for military purposes. Other countries, such as the United States and Israel, have set a worrying precedent, so there is some fear that these technologies could be misused and end up being lethal.
The steps of China. China has long stressed the importance of AI always having human supervision. However, the announcement that have collected media such as Interesting Engineering follows a different premise. The Asian country intends for its satellite AI systems to be capable of breaking down complex tasks, coordinating workflows and recovering independently from possible failures.
In short, they would use algorithms capable of analyzing information, making decisions and acting without the need for humans. Satellite observation can have many uses, from analyzing animal behavior to aiding weather forecasts. However, this emphasis on AI being able to act on its own is inevitably reminiscent of what we are already seeing in the United States for military purposes.
The case of the United States. The US military is suspected of having several AI-based targeting systems, although there is a lot of classified information about them. Broadly speaking, these would use data obtained through optical cameras, infrared, radar, LIDAR and other sensors to locate military targets and subsequently shoot if deemed necessary. Last February, the attack on a school classified as erroneous opened the debate on this issue. In it 175 people diedmostly girls. As it was recognized as a misguided attack, suspicions grew that AI was behind it. As this is classified information, it cannot be known for sure, but it is a more than tangible risk.
USA has hired currently to SpaceX to improve the connectivity of the shooters so that they receive the information as soon as possible whether to shoot or not. It is clear that they want to continue advancing in this aspect and it is something that, logically, is of great concern to experts.
The case of Israel. It is not exactly the same, but Israel has also left war decisions in the hands of AI, with worrying results. For example, they have a system that uses data from phone calls, social networks, metadata, visual information or contacts, among other sources, to determine who may be targets. According to an investigation by +972 Magazinein the first weeks of the war against Palestine detected 37,000 targets as Hamas members. The army itself recognized that the algorithm has a 90% probability of being right. This gives us clues about how catastrophic it can be.
More transparency? China has assured that it will have more transparency than other countries by introducing AI in the management of its satellites. However, this does not minimize the concerns of experts. For now, it seems that the algorithms have managed to avoid obstacles independently in the testing phase. For them to be able to shoot, if they are used for that purpose, there is still a long way to go. But it is viable. Therefore, it is urgent that measures be taken to regulate the use of AI without human supervision as much as possible.
There are too many human beings with few scruples, but even those may have some more qualms than machines. An AI cannot show concern, conscience or ethics. He also does not ask before shooting if he is not told to do so. Therefore, if we want it to operate satellites capable of observing and controlling what we do on the planet, it would be advisable for us to ensure that someone with scruples remains in charge. Although in some contexts that is difficult to find.
Image | Kevin Stadnyk (Unsplash)/Magnific
In Xataka | There are planes dropping food into Gaza from the air. It is a worse idea than it seems to fight hunger
GIPHY App Key not set. Please check settings