Updated Estimates on Simulation Probability

In the latest edition of my 2025 book, “Simulation Hypothesis”, we’ve revised our estimate regarding the likelihood of existing within a simulation to around 70%. This insight stems from recent advancements in AI technology. Essentially, it’s becoming increasingly plausible that we might be inhabiting a virtual reality along the lines of the scenarios portrayed in the iconic film, The Matrix, released in 1999.

Interestingly, even those who weren’t born when The Matrix hit the screens seem to grasp its essence. Neo (played by Keanu Reeves), for example, is a character convinced he’s living a real life while working for a colossal software company.

Once you achieve the simulation threshold, it opens up the possibility to design your own “matrix” full of lifelike landscapes and AI characters.

Why the Certainty About Simulations?

This book dives deep into various explanations surrounding this notion. It analyzes quantum peculiarities, time and space’s oddities, information theory, and even philosophical and spiritual viewpoints, which attempt to address the defects within a simulated matrix.

AI’s Role

Even if we were to simulate these possible explanations, the main driver for our new estimates is largely due to the impressive acceleration in AI and virtual reality advancements. Notably, Nick Bostrom’s philosophical ideas proposed back in 2003 about statistical simulation scenarios have taken on new meaning with the rise of generative AI platforms like ChatGPT and Google Gemini. We now interact with AI entities that not only pass the Turing test but also exist within a virtual space.

A recent illustration of this is Google’s new Veo feature. They’ve rolled out a tool that produces convincing videos on demand, featuring virtual actors delivering lines based on user prompts. This led to an amusing situation where people speculated about an AI-generated video where characters insist they are not, in fact, AI creations.

The Surge of Virtual Companions

Another recent advancement includes the launch of Grok’s AI companion, merging large language models with virtual avatars. This represents a significant leap forward in the world of virtual interactions, where AI characters assume roles like friends, therapists, and even educators. There’s been a notable obsession with particularly enticing virtual characters, showcasing an evolution in visual fidelity and responsiveness—just think of the realism in Google Veo videos paired with virtual companions.

In this context, we seem closer than ever to reaching what is termed “simulation points.” This concept describes a theoretical stage enabling the creation of virtual worlds indistinguishable from physical reality. Once this threshold is crossed, one can craft a “matrix” that feels genuine.

Exploring Ancestral Simulation

To grasp why advancements toward this point may suggest we could already be part of a simulation, we delve into Bostrom’s argument from his seminal 2003 paper, “Do You Live in Computer Simulations?”.

Bostrom proposed three potential scenarios regarding the creation of realistic simulations by technologically advanced civilizations. Each simulation would contain minds that mimic the capabilities of biological brains. These scenarios mirror the concept of “simulation points.”

The first two outcomes could suggest that no civilization reaches a simulation point—either due to self-destruction or inability—or that civilizations capable of creating such detailed simulations choose not to do so.

The term “simulation hypothesis,” as initially posited by Bostrom, indicates the third scenario: “We are almost certainly living in computer simulations.” The reasoning here is that advanced civilizations would produce simulated worlds at the touch of a button, thus multiplying the number of simulated beings far beyond biological ones. Statistically, it would imply that if one could not tell the difference, they would be more likely to be a simulated being than a biological one.

With advancements in today’s AI, we estimate that at least 67% of chances to reach simulation points are likely, likely even exceeding 70%.

Bostrom initially refrained from assigning a percentage to this third scenario, suggesting a one in three chance regarding the potential outcomes—not a clear-cut conclusion. Elon Musk later adjusted this reasoning in 2016, suggesting there are likely billions of simulated worlds, making the chance of living in the actual reality minimal by comparison.

What Are the Odds?

Others have also chimed in on this debate. For instance, Neil deGrasse Tyson proposed that the odds could be around 50%. Colombian scientist David Kipping, using Bayesian methods and Bostrom’s ideas, reached similarly low estimates.

Musk focused on the rapid evolution of video games, and I elaborate on this in my book. It lays out ten stages that would lead to simulation points, covering technologies like Virtual Reality (VR), Augmented Reality (AR), Brain-Computer Interfaces (BCIs), and AI. The progress in these areas bolsters the belief that we are inching closer to such simulation points.

I argue in my book that the likelihood of us being in a simulation mostly hinges on our ability to reach a simulation point. If we can’t, then the probability drops significantly. Achieving that point, however, could mean being 99% certain we are participating in a simulation. Even if we haven’t reached it yet, the potential probability essentially condenses to our confidence in attaining this threshold, adjusted for some uncertainty factors.

If we believe we can reach these simulation points, the higher likelihood of being within an existing simulation grows. The implications suggest an active development of a “matrix” by someone, heightening the probability that we might already be experiencing it.