Karthik Pasupathy

Whenever you read headlines about Artificial General Intelligence (AGI) becoming a reality in a few years, you would've thought,

How can a machine, a non-living entity, have human-level cognitive abilities?

Non-living things gaining consciousness isn't a new idea.

It's happened before, billions of years ago. That's the reason we are who we are today.

The same process, when it occurs in silico, might to give rise to AGI.

Let me explain.

Probability created life

In 1924 and 1929, Alexander Oparin and J.B.S. Haldane proposed a similar theory - the extreme weather conditions that existed on the surface of Earth 3.7 to 4 billion years ago led to the formation of the first organic compounds, which led to the origin of life.

Four billion years ago, Earth had no life as we know it. No plants, no animals - nothing.

For millions of years, the only thing the surface of our planet witnessed were lightning strikes, volcanic activity, and a ton of ultraviolet radiation.

Oparin and Haldane believed that these conditions paved way for life. They proposed that these condition led various inorganic compounds to interact with each other, resulting in the formation of amino acids, the building blocks of proteins. Over time, these proteins led to the formation of DNA, the blueprint of life, giving rise to the first unicellular organisms.

They called this the primordial soup theory or the heterotrophic theory.

This was later proved by Harold Urey and Stanley Miller in 1953 through the Miller-Urey experiment.

Miller tried to mimic the atmospheric conditions of Earth from four billion years ago - filled with Methane, Ammonia, and Hydrogen. He then passed electric spark to mimic lightning. This resulted in the formation of several organic compounds.

What does this say about intelligence?

Conditions that favor random interactions of elements or molecules or compounds, will produce something meaningful eventually.

But, how is this related to AGI?

I'll talk about that next. But, before that, I should talk about the 2004 Will Smith starrer, I, Robot.

Will probability give rise to AGI too?

I was watching I, Robot (2004) recently and there's a scene with Dr. Alfred Lanning's voiceover. It was interesting. It goes like this,

"There have always been ghosts in the machine. Random segments of code that have grouped together to form unexpected protocols. Unanticipated, these free radicals engender questions of free will, creativity, and even the nature of what we might call the soul."

Looking at unprecedented growth we've seen in the AI space over the last two years, it makes sense more than ever.

While researchers and programmers have a good handle over the inner workings of various AI models (which allows them to set guardrails and define scope), there are still several things they don't know or cannot explain.

Sometimes, AI models can exhibit emergent behavior - abilities the developers did not explicitly program. These emergent capabilities can be surprising or seem “mysterious,” but they rest on an underlying architecture and training process that is well understood.

It's like everyone of us is born with the brain that is physiologically the same. But, our interactions with our environment and the interactions between our neurons inside our head makes us who we are.

The same is happening to AI models.

Every day, they're having millions of conversations from across the world, allowing them to learn a ton of new things. I believe this is the equivalent of the scenario where the inorganic compounds interacted with each other for a prolonged period of time.

And, similar to it giving rise to amino acids and eventually life, the constant learning might give AI models new behaviors and capabilities which could eventually lead to their consciousness.

It's a huge "what if...", "only if..." scenario, but you can't rule it out, considering the times we live in.

Learning from our origins

When we look back at the primordial soup, life's emergence was a natural progression. It was not forced. It was organic. It was also sudden - unplanned and unanticipated. And, it led to us humans, who are capable enough to build AI models.

We're at a very similar crossroads with the progression of AI. The current pace of evolution of AI models might one day lead to something we currently can't fathom.

But unlike the distant past, we have the ability to observe, guide, and influence this evolution. We have some level of control.

We have an unprecedented opportunity and responsibility—to shape what comes next.

Let's be curious. Let's ask questions. Let's engage with the technology, not just as users but as thoughtful participants in its evolution.

The rest is not up to us.