Physicist believes that the entire universe could be a neural network.

(Image credit: Shutterstock)

We do not often encounter research that attempts to change the laws of reality.

However, in a preprint released on arXiv this summer, Vitaly Vanchurin, a physics professor at the University of Minnesota Duluth, made an especially startling attempt to reframe reality by proposing that we live inside a massive neural network that controls everything around us.

The paper cited a "possibility that the entire cosmos on its most fundamental level is a neural network," to put it another way.

For years, physicists have been trying to make sense of the conflict between general relativity and quantum physics. While the second argues that time is variable and entwined with the structure of space-time, the first maintains that time is universal and absolute.

According to Vanchurin's article, artificial neural networks are able to "display approximate behaviours" of both universal theories.  "It is widely believed that on the most fundamental level, the entire universe is governed by the rules of quantum mechanics, and even gravity should somehow emerge from it," according to him, as quantum mechanics "is a remarkably successful paradigm for modelling physical phenomena on a wide range of scales."

According to the paper's commentary, "We are not merely claiming that artificial neural networks can be beneficial for evaluating physical systems or uncovering physical laws; we are saying that this is how the world around us actually works." "In this regard, it may be viewed as a proposal for a theory of everything, and as such, it should be simple to disprove."

The majority of machine learning specialists and physicists we contacted said they were skeptical of the paper's findings and refused to comment on the record. Vanchurin, however, explored the argument in further detail and shared more details about his idea in a Q&A with Futurism.

Futurism:  According to your paper, the universe may be essentially a neural network.  How would you defend your position to someone who was not familiar with physics or neural networks?

Vitaly Vanchurin: There are two ways to answer your question.

The first method involves creating a thorough model of neural networks first, and then examining how the network responds to a large number of neurons.  I have demonstrated that the behavior of a system close to equilibrium is accurately described by quantum mechanics equations, whereas the behavior of a system farther from equilibrium is accurately described by classical mechanics equations.  Is it a coincidence?  Perhaps, but as far as we know, the physical world operates according to both quantum and classical mechanics.

Starting with physics is the second choice. Although general relativity and quantum mechanics are known to function well on large scales and small scales, respectively, we have not yet been able to integrate the two theories into a cohesive whole. This is referred to as the problem of quantum gravity. There is obviously a big gap, but to make it worse, we don't know how to handle observers. This is referred to as the measurement problem in quantum physics and the measure problem in cosmology.

Then, it may be argued that three phenomena—rather than two—need to be brought together: general relativity, quantum physics, and observers. 99 percent of physicists believe that quantum mechanics is the primary one and that all else should flow from it in some way, but nobody knows how. Another option that I put up in this study is that everything else, including general relativity, quantum mechanics, and macroscopic observers, arises from a microscopic neural network. Everything seems to be going smoothly so far.

What first gave you this idea?

For first, I just wanted to understand more about deep learning, thus I prepared a paper called "Towards a Theory of Machine Learning."  Under some restrictions, the learning (or training) dynamics of neural networks prove to be very comparable to quantum dynamics observed in physics, despite the initial plan to analyze the behavior of neural networks using statistical mechanics techniques.  The idea that the physical world is basically a brain network caught my attention while I was on sabbatical leave, and I am still on it now.  The idea is wild, but is it wild enough to be real?  We'll have to wait and see.

"All that is needed to verify the theory is to uncover a physical phenomenon which cannot be explained by neural networks," you wrote in the article. What are you referring to specifically? Why is this situation "easier said than done"?

The great majority of the "theories of everything," which are numerous, must be wrong.  Finding a phenomenon that cannot be explained by a neural network is all that is required to refute my theory, which holds that everything you see around you is a neural network.  But when you think about it, it's a really challenging task since we don't really understand how machine learning and neural networks work.  That is the reason I initially tried to develop a machine learning theory.

 Although the idea is crazy, is it crazy enough to be real?  We'll have to wait and find out.

Reference