We don't come across studies that attempt to change reality's laws every day.
But in a preprint posted on arXiv this summer, a physics professor at the University of Minnesota Duluth by the name of Vitaly Vanchurin tried to reframe reality in a particularly eye-opening way, contending that humans are governed by a vast neural network while living inside of one.
He wrote in the article that it's "possible
that the entire cosmos is a neural network at its most fundamental level."
For many years, physicists have been working to make quantum physics and general relativity compatible. While the second maintains that time is relative and entangled with space-time, the first maintains that time is universal and unchanging.
In his article,
Vanchurin asserts that artificial neural networks are capable of
"displaying approximate behaviours" of both universal theories. Because
quantum mechanics "is a remarkably successful paradigm for modelling
physical phenomena on a wide range of scales," he writes, "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."
"We are not only asserting that artificial neural networks can be helpful for assessing physical systems or identifying physical rules; we are saying that this is how the world around us actually works," the paper's commentary reads. It may be seen as a proposal for a theory of everything in this sense, and as such, it should be easy to refute.
We contacted a number of physicists and machine learning specialists, but the majority of them declined to speak with us about the paper's findings on the record. However, Vanchurin went into further detail regarding the argument and his concept during a Q&A with Futurism.
Futurism: According to your paper, the universe might be fundamentally a neural network. How would you justify your conclusions to someone who was unfamiliar with physics or neural networks?
Theodore Vanchurin: Your question might be answered in two different ways.
The first method involves building a thorough model of a neural network and then analysing how the network functions with a large number of neurons. I've demonstrated that whereas classical mechanics equations accurately describe the behaviour of a system that is further from equilibrium, quantum mechanics equations accurately describe the behaviour of a system that is near equilibrium. Coincidence? Maybe, but as far as we are aware, the physical world operates according to quantum and classical mechanics.
Starting with physics is the second possibility. We are aware that general relativity is effective at large scales and quantum mechanics is effective at small scales, but we have not yet discovered a mechanism to integrate the two theories. The quantum gravity problem refers to this. It's obvious that a crucial component is missing, but to make matters worse, we have no idea how to handle observers. This is referred to as the measurement problem in cosmology and quantum mechanics, respectively.
Then, rather than two phenomena, it may be argued that three phenomena—quantum mechanics, general relativity, and observers—need to be combined. Quantum mechanics is the primary one, according to 99 per cent of physicists, and everything else should follow from it in some way, but no one knows how that can be accomplished. In this paper, I suggest a second possibility: that everything else, including quantum mechanics, general relativity, and macroscopic observers, originates from a microscopic neural network. Everything seems to be going fine so far.
Who or what initially thought of this?
I first produced a paper titled "Towards a
Theory of Machine Learning" because I merely wanted to understand more
about deep learning. The initial intention was to analyse the behaviour of
neural networks using statistical mechanics techniques, but it turned out that,
within certain bounds, the learning (or training) dynamics of neural networks
are strikingly comparable to quantum dynamics encountered in physics. I made
the decision to look into the idea that the physical universe is really a
neural network while I was on sabbatical leave at the time (and still am). The
idea is absurd, but is it really absurd enough to be true? We'll have to wait
and see about that.
You wrote in the study, "All that is needed to prove the theory is to identify a physical phenomenon which cannot be described by neural networks." What do you mean specifically? Why is it in this situation "easier said than done"?
The vast majority of the "theories of everything" must be false because there are so many of them. My hypothesis is that everything you see around you is a neural network, so all it takes to refute it is to identify a phenomenon that cannot be explained by a neural network. But when you think about it, it's a very challenging task since we don't fully understand how neural networks and machine learning operate. That is the main reason I initially tried to develop a machine learning theory.
The idea is absurd, but is it really absurd enough
to be true? We'll have to wait and see about that.
Reference: ArXiv
No comments