We have long been under the impression that, since we created artificial intelligence, we are able to control it. However, a new study just revealed that there are limits to our ability to control AI.
According to this new study that came out of the Max-Planck Institute for Humans and Machines, humans were incapable of stopping an AI machine from predicting what decisions it might make, and then making its own decisions. Some of these machines have begun to perform tasks that we can’t understand how they learned, and now, researchers have found that certain algorithms containing AI have proven to be incomputable—which means that they are unable to be calculated or estimated by humans.
Though the idea of a human-built machine that humans don’t understand and can’t control may be unsettling to some, this technology is already here. If we learned anything from Netflix’s The Social Dilemma, it’s that algorithms have taken on a life of their own, and we may no longer have control over artificial intelligence.
This very much sounds like a “you made your bed now lay in it” scenario, and for the most part, it is. This study’s co-author, Manuel Cebrian, explains in a statement: “There are already machines that perform certain important tasks independently without programmers fully understanding how they learned it. The question therefore arises whether this could at some point become uncontrollable and dangerous for humanity.”
What we are beginning to learn is that we may not be able to predict or stop super-intelligent machines from evolving. We have been previously warned of the nefarious potentials of powerful AI, but it seems that we have to endure the consequences now or later.