Advancing Artificial Intelligence Using Quantum Mechanics
As the tasks which we entrust to artificial intelligence become increasingly important and complex, the ability to program adequate decision-making skills will be an absolute necessity. Robots will need to have the ability to autonomously react to their environments and make decisions based on their past experiences. Now, researchers from the Austrian Academy of Sciences, the University of Innsbruck and the Complutense University in Madrid have incorporated quantum mechanical principles into their programming in order to speed up robots' decision-making processes.
"Our agent model comes to a rational conclusion faster because, by using quantum physics, it is, in a manner of speaking, able to recall previous experiences simultaneously, searching for the best action," said team leader Hans Briegel.
Modern artificial intelligence has a robotic version of episodic memory that consists of a network of "clips," or data from the robots' past experiences. When the AI is faced with a new situation, it decides on a reaction by performing a randomized search of its clips and synthesizing a combination of different clips that functions like a new memory. Or, more analogously, it functions as an imagined projection into a situation, and the AI then decides between the various projections. The research team aims to incorporate quantum mechanics into this process by replacing the randomized walk with a non-deterministic (stochastic) quantum process, making it quadratically (comparable to exponentially) faster: "A random walk is substituted by a quantum stochastic process, which allows for a more efficient exploration of the memorized experiences," explains junior scientist Vedran Dunjko. "This is why the quantum agent is considerably faster or to be more precise, quadratically faster in terms of taking action than a conventional agent."
Although the researchers acknowledge that this research is in its infancy, Briegel is confident that "studying and using autonomous and active learning systems in the context of quantum experiments could steer research into entirely new and exciting directions." This technology may not only allow an AI to react to stimuli more quickly and be adaptable in quickly-changing environments, but may also allow an AI to undergo the process of learning much as a human would. Quick "reflexes" will be crucial to many of the functions that we will eventually expect robots to be able to perform, particularly technology such as self-driving cars.