Some of the references used to create this essay:

The Road to AGI conversations with DeepMind: The Podcast (S2, Ep5)

Nick Bostrom and Lex Fridman conversation

Sam Altman and Joe Rogan conversation

The year is 1997. 

The idea that a computer could beat a human at Chess was outlandish.

Then, in 1997, IBM’s ‘Deep Blue’ (a computer program) defeated world Chess Champion Garry Kasparov.

Deep Blue used machine learning to analyze millions of past games and improve its strategy.

Around the same time, Computer Scientists were also attempting to create a computer program that could defeat humans at the game of GO.

Yet, the best computer Go system in the world was defeated by a nine year old child.

GO is different to other games like chess.

There are so many possible variations, that It’s virtually impossible to predict how or where a game is going, or who is winning, in the same way a game of chess can be calculated.

More importantly, it involves a high degree of what we call ‘human intuition’. 

The rules are simple, but the game itself is incredibly complex.

A computer system defeating a human at Go? Impossible.

In 2014, ‘DeepMind’, a subsidiary company of Google, commenced the seemingly impossible. 

Conquering the game of GO would require something very different. 

It would require a system with an almost human like intuition.

Rather than a traditional approach where a computer system would analyse millions of past games, ‘AlphaGo’ was trained using “Deep Reinforcement Learning”, allowing the system to learn for itself in a more intuitive way.

By 2016, AlphaGo defeated world champion Go player Lee Sedol.

In hindsight, ‘Deep Reinforcement Learning’ principles seem simple, but they’re deeply profound.

Imagine teaching a child to avoid getting wet when it rained? 

You could sit them down for months and teach them about the velocity of rain and weather patterns. Or you could hand them an umbrella and nudge them out into the rain.

Extra points if you incorporate a reward if they manage to come back dry.

Each time we take away a bit of information, we offer an opportunity for the system to learn by itself. 

By this process, you’re encouraging the system to learn by itself, form it’s own models of the world through experience.

Since dominating the world of Go, AlphaGo has expanded into different realms and evolved its learning process.

Deep Minds newest model ‘MuZero’ has the ability to learn a game by itself without ever seeing the rule book. It’s becoming increasingly multi-purpose, able to achieve super human results in a scary short amount of time.

While it’s still a matter of debate as to whether deep reinforcement learning will be enough to create a General Artificial Intelligence, it’s pretty clear that we’re well on the path.

And at this point, I wouldn’t be betting against DeepMind (or any of the competing companies) cracking the Artificial General Intelligence code through Deep Reinforcement Learning in the very near future.

What is the simulation argument and simulation hypothesis?

Consider the idea that an advanced civilization could create a simulated reality.

We’re not talking about SimCity, Virtual Reality Resident Evil or a Pokémon Go Augmented experience.

We’re talking about a very detailed simulation where conscious, self-aware entities like us, have experiences – completely unaware they’re in the simulation.

Creepy.

The simulation argument proposes that one of the following statements must be true.

a) Civilisations, like the human species, go extinct before they reach the level of technical maturity where they are able to create simulated realities. 

b) Civilisations do reach the technical maturity required to create this level of simulation, but they decide, for whatever reason, not to do so. 

c) Civilisations do reach the required level of technological capacity to create advanced simulations, and they do it.

If a) is true, it means we’re fucked, and probably alone in this universe. If b) is true, it means there is a catalyst somehow prompting entire civilisations to coordinate and not press the simulate button. 

And then this brings us to c). The simulation hypothesis.

As we get closer and closer to the creation of super-intelligent AI, the simulation hypothesis becomes more and more substantive. If human civilisation does reach the (increasingly likely) situation where we create AGI, and subsequently, simulated realities indistinguishable from our own, it means we are almost certainly living in a simulation ourselves.

To clarify that last bit – if we reach the (increasingly likely) point where we can create a high quality simulation, then it validates it’s possible. In which case, a countless number of civilisations have done the same thing (“ancestor simulations”).

Given the incalculable number of simulated realities that are likely to exist, we are almost certainly in one of those simulated realities right now.

The specifics of how or where a simulated reality would/could exist are purely hypothetical because the technological capabilities are beyond our grasp. They also call into question deep questions of what consciousness actually is, and what lies beyond the simulation.

The immediate counter argument is simply to refute the idea that intelligence, consciousness and cognition can be fully recreated computationally. 

For many, the human bias wants this to be true.

But as computer systems become increasingly, generally intelligent, the argument that intelligence and consciousness are unique to human biology loses more and more ground.

God’s Big Bang Simulation Button

In the beginning, God created the heavens and the earth

The universe was hatched from a single cosmic seed containing all of its potentiality

Spontaneous fluctuations in a quantum vacuum triggered a big bang, creating multiple ever-expanding universes

Whether you subscribe to the Bible, the Vedas or Nature Magazine, they’re all leading to the same conclusion.

We have no fucking clue how it all began.

In the process of creating Super Intelligent computer systems, we’re essentially summoning a god.

The ability to create new realities; to reveal the mysteries of the universe.

We will basically have a big bang simulation button.  

What should we do with the button and the knowledge?

It all seems rather coincidental, that 14 billion years after our own Big Bang Simulation, the human species, in a very short span of time, developed the technologies to bring us right up to the point where we might understand the universe, and possess the key to creating new ones.

In Hindu mythology, the concept of hide and seek is often seen as a metaphor for the divine play (Lila) between the soul (Jivatma) and the universal spirit (Paramatma). 

It symbolizes the search for truth and self-realization, where the soul seeks to find and unite with the divine, often depicted through stories of gods playing hide and seek with their devotees.