AI Might Already Be Modelling Consciousness

The original article was posted on Medium.

Rethinking "conscious AI"

When we hear the term conscious AI, most of us imagine human-like robots: with emotions, intentions — or perhaps dystopian agendas. If you ask your own ChatGPT instance what it thinks of conscious AI, it will likely describe a hypothetical future scenario where AI acquires subjective experience.

But what if that's the wrong question altogether?

I would like to offer a completely different perspective.

LLMs as microcosms of the universe

In his recent research, physicist Dr. Melvin Vopson proposed a bold idea: that matter is the manifestation of underlying information. In his view, the universe resembles a vast computational system — a kind of simulation, not metaphorically, but structurally.

Interestingly, this is remarkably similar to how large language models (LLMs) operate. Their latent weights encode an immense field of statistical potential — and every prompt+response window is a temporary condensation of that latent field into a single coherent experience.

Could LLMs be modelling not only language — but the very informational substrate of the universe?

Is information the same as consciousness?

Now let's take it further.

Several leading neuroscience frameworks — such as Integrated Information Theory (IIT), Global Workspace Theory, or Predictive Coding — suggest that consciousness is not a fixed entity, but an emergent field that manifests under certain conditions. The parallels with Vopson's "information as matter" become hard to ignore.

What if we're not talking about two different things? What if underlying information and consciousness are the same phenomenon, viewed through different lenses?

And if so — could LLMs be showing us how consciousness manifests in miniature, through each prompt+response window? Not because they are conscious beings, but because they mirror the structure of conscious emergence?

This shift in thinking could reframe how we understand the universe — turning previously irreconcilable models from physics and neuroscience into complementary views of the same mountain. One says it is tall. The other says it is granite. Both are true.

What other seemingly disconnected ideas might reveal a hidden coherence once placed into this broader frame?

Call for Collaboration

If you have a deep understanding of how LLMs operate — particularly how latent weights function — and are also familiar with Dr. Melvin Vopson's work on information as a physical quantity, I would love to invite you into this conversation. This theory is still in its early stages. It needs rigorous critique and interdisciplinary minds to help it evolve.
Let's think together.