As artificial intelligence (AI) continues to evolve and improve, researchers are offering a dire warning, saying super-intelligent AI will be impossible to control.

AI is one of the most controversial technological developments. Its proponents claim it will revolutionize industries, solve a slew of the toughest problems and lead to the betterment of humankind. Its critics believe it represents an existential threat to humanity, and will eventually evolve beyond man’s ability to control it.

An international team of researchers are now saying AI will evolve beyond our ability to control it, based on theoretical calculations. In a paper published in the Journal of Artificial Intelligence Research, researchers Manuel Alfonseca, Manuel Cebrian, Antonio Fernandez Anta, Lorenzo Coviello, Andrés Abeliuk and Iyad Rahwan make the case “that total containment is, in principle, impossible, due to fundamental limits inherent to computing itself.”

While it may seem unlikely that AI could evolve to such a point, co-author Manuel Cebrian, Leader of the Digital Mobilization Group at the Center for Humans and Machines, Max Planck Institute for Human Development, argues that AI is already reaching this point to some degree.

A super-intelligent machine that controls the world sounds like science fiction. But 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.

The study, entitled “Superintelligence Cannot be Contained: Lessons from Computability Theory,” could very well have far-reaching implications for AI research.