Physicist Proposes We Might Be Living Inside a Giant Neural Network

Rarely, do we encounter a document that aims to alter the nature of reality as we understand it.

Yet in this summer’s hotly discussed preprint on arXiv, a professor of physics at the University of Minnesota Duluth, Vitaly Vanchurin, tries to explain the nature of reality in a rather revealing way, asserting that we are inside the gigantic neural network that rules all around us. In other words, he stated in the report that it is conceivable that the entire universe at root “is a neural network. ”

Currently, there has been a massive effort by physicists to combine quantum physics with general relativity. The one postulates that time is the eternal, real, and continuous, while the later argues that time is relative and a function of the space time continuum.

According to the findings of Vanchurin, artificial neural networks can work with approximate behaviors of both universal theories. Since, as he notes, “quantum mechanics is a remarkably successful paradigm for modeling physical phenomena on a wide range of scales” so there is a common and rather popular opinion that “on the most fundamental level the entire universe is described by the rules of quantum mechanics and even gravity should somehow derive from it. ”

“Rather than insisting that the artificial neural networks may be useful for the study of physical systems or for identifying physical laws, it is affirming that this is how the physical world operates”, the paper’s discussion suggests. “In this respect, it could be considered as a proposal of the theory of everything, and as such it should be quite easy to disprove. ”

The idea is so provocative that the majority of physicists and machine learning specialists that we interviewed refused to be interviewed on the record citing the concerns about the outcomes of the paper. But in the Q&A session with Futurism, Vanchurin decided to go deeper into the topic – and share more about his idea.

Futurism: Your paper contends that the cosmos may be fundamentally a neural network. How would you communicate your logic to someone who knows nothing about neural networks or physics?

Vitaliy Vanchurin: There are two possible answers to your inquiry.

The first approach is to begin with a probe model of the neural networks and then analyze the characteristic of the network as the number of neurons goes to infinity. What I have been able to demonstrate is that equations of quantum mechanics are quite capable of predicting the behavior of system close to equilibrium and that equations of classical mechanics are quite capable of predicting the behavior of a system farther away from equilibrium. Coincidence?maybe, but as for now quantum and classical mechanics depict how the physical reality works and is expected to work.

The second approach that can be taken is to begin with physics. Quantum mechanics is known to apply well on small dimensions while relativity applies well on large dimensions and there is no clear link between the two. This is known as the “quantum gravity problem. ” Obviously, there is something fundamental we are not seeing; this is made even worse by the fact that there seems to be no way of handling spectators. This is referred to as the measurement problem in quantum mechanics and the measure problem in cosmology, respectively.

The second option suggests beginning with the subject of physics. They both are very good on their regimes with quantum mechanics on small sizes and general relativity on big sizes we don’t have a theory that combines the two. This situation is called the “quantum gravity problem. ” It seems obvious that we are failing to apprehend something significant, and, what is more, we have no idea how to address spectators. This is referred to as the measurement problem in quantum mechanics and the measure problem in cosmology, respectively.

Then one may argue that three phenomena must be unified: quantum mechanics, general relativity and observers. If physicists were to rate the importance of theories, 99% of them would say quantum mechanics is the most important one and that everything else should be derived from it; but nobody knows how in detail. In this study, I present another possibility: a given micronetwork is the primary architecture from which everything else unfolds, including quantum mechanics, general relativity, and macroscopic observers. So far, all the signs point to the positive.

What prompted you to come up with this idea?

First, I wanted to know more about how deep learning operates and thus I wrote a paper titled “Towards a theory of machine learning“. The aim was simple: apply statistical mechanics techniques to studying neural network behavior, but it became clear that under some restrictions, the learning process of neural networks resembles quantum processes discovered in physics. At the time of writing, I was (and remain) on sabbatical, so I chose to look into the theory that the physical universe is a neural network. Of course, the concept is rather peculiar but that does not necessarily mean that the idea is true. Wait and see.

In the publication you said that to prove the theory was wrong, “all that is needed is to find a physical phenomenon which cannot be described by neural networks. ” What exactly do you mean by that?That is why it is said that ‘it is easier said than done’.

Well there are many theories of everything and probably the vast majority of them have to be wrong. In my hypothesis, everything you see around you is a neural network and hence to prove it wrong, all that is needed is to come up with a phenomenon which cannot be translated wit a neural network. But if you really consider it that is a pretty challenging task partly because we are not exactly sure how the neural network is functioning and how the machine learning is functioning. That was why I attempted to formulate a theory of machine learning on the first place.

The idea is definitely crazy, but if it is crazy enough to be true? That remains to be seen.

How does your study connect to quantum physics, and does it deal with the observer effect?

There are two main schools of thinking on quantum mechanics: Many-worlds interpretation – Everett and Hidden variables – Bohm. I don’t think I have anything fresh to add to the discussion regarding the many-worlds interpretation, but I think I can contribute something to hidden variable theories. In the emergent quantum mechanics model for the neural network I reviewed, the hidden variables are the states of the neurons while trainable variables including the bias vector and the weight matrix are quantum variables. Therefore, it should be noted that the hidden variables can be highly non-local, which will violate Bell’s inequalities. An approximation space-time locality is expected to emerge, but fundamentally speaking, each neuron can be connected to any other neuron, and therefore the system can be nonlocal.

Would you mind expanding on how this idea connects to natural selection? How does natural selection influence the evolution of complex structures/biological cells?

What I’m saying is really basic You’re probably better off not having been born if, on the other hand, you are going to spend your life spouting idiotic theories like mine. Some of the structures in the micronetwork, which can be termed as subnetworks, are in stable forms while others are in unstable forms. It also posited that while the more stable structures would continue to be sustained through the evolutionary process, the less stable structures would be eliminated. At the lowest sizes I hypothesize that the process of this natural selection should produce least complex structures like chains of neurons, but in higher dimensions the structures are more complex. As I don’t observe any reasons why this method can’t operate at certain L scale, the argument is that all what we observe around (particles, atoms, cells, observers, etc. ) is a result of natural selection.

I was intrigued by your first email when you said you might not understand everything yourself. What did you mean by that? Were you referring to the complexity of the neural network itself, or to something more philosophical?

Yes, I only refer to the complexity of neural networks. I did not even have time to think about what could be philosophical implications of the results.

I need to ask: would this theory mean we’re living in a simulation?