Chaos Doesn’t Care About Your Substrate. Consciousness, AI, and the Mess That Makes Us Alive

A Boring Book That Made Me Think

I was 42 minutes from finishing Feeling & Knowing by Antonio Damasio when something clicked. The book is dense. Academic. At times, punishingly dry. But underneath the neuroscience jargon is an idea that quietly touches on what’s happening right now with artificial intelligence.

Damasio’s argument is this: consciousness didn’t appear out of thin air as some mystical gift from the universe. It evolved. Gradually. From the body’s need to not die.

That’s it. That’s the whole book. The body has to regulate itself, maintain temperature, chemistry, structure, or it stops existing. Damasio calls this homeostasis. And he argues that feelings are the mind’s way of monitoring that process. Pain means something is wrong. Pleasure means something is right. Fear means something might kill you. Comfort means you’re safe, for now.

Consciousness, in his framework, is what happens when a system gets complex enough to know that it’s feeling. Not just react. Not just adjust. But experience the adjustment. A “self” emerges that owns the sensation.

Being. Feeling. Knowing. Three layers, built on top of each other over billions of years of evolution. And all of it starts with one thing: an organism that has something to lose.

• •  •

The Goal That Started Everything

Before there was feeling, before there was knowing, there was a goal. The simplest goal any living thing can have: survive.

A single-celled organism doesn’t think. It doesn’t feel. But it moves toward nutrients and away from toxins. It has a goal baked into its chemistry, stay alive long enough to replicate. That’s not consciousness. But it’s the seed of it.

Over millions of years, organisms that were better at pursuing that goal, better at sensing threats, finding resources, avoiding danger, survived. The ones that weren’t, didn’t. And as environments became more complex, the internal systems required to navigate them became more complex too. Simple chemical reactions became nervous systems. Nervous systems developed the ability to monitor internal states. Internal monitoring became feeling. Feeling, eventually, became awareness.

Consciousness didn’t appear because the universe wanted it to. It appeared because survival demanded it. The goal came first. The awareness came after, as a tool to serve the goal.

• •  •

Consciousness Was Forged in Chaos

But survival against what? That’s the part worth paying attention to. The reason consciousness exists is because life is an absolute mess.

Think about what a human being processes in a single day. Not computes, processes. The alarm goes off and you’re already managing competing signals: exhaustion says stay in bed, responsibility says get up, anxiety says you’re already behind. You haven’t even opened your eyes yet and your consciousness is negotiating a three-way conflict between your body, your obligations, and your fears.

Then the day actually starts.

You navigate traffic with people who are distracted, angry, or incompetent. You manage relationships with colleagues who have their own agendas, insecurities, and bad days. You make decisions with incomplete information under time pressure. You love people who can hurt you. You trust people who might betray you. You build things that might fail. You invest years into things that might not matter.

And underneath all of it, running constantly, is the quiet hum of mortality. The awareness that this is finite. That every hour spent is an hour you don’t get back. That the people you love will leave or be taken. That the body carrying your consciousness is degrading in real time, and one day it will stop.

Human consciousness isn’t a clean operating system. It’s a survival tool forged in fire.

We love and we betray. We create and we destroy. We know exactly what we should do and choose not to do it. We lie to ourselves about why we made decisions. We carry grudges that serve no purpose. We chase status instead of substance. We procrastinate on the things that matter and obsess over things that don’t.

This isn’t a flaw in consciousness. This is the environment consciousness was built to navigate. Every contradiction, every competing drive, every irrational impulse, that’s the chaos. And consciousness is what emerged because some organism, millions of years ago, needed a way to make sense of a world that made no sense.

• •  •

Two Opposing Ideas

There’s a prevailing view in neuroscience that consciousness requires a body. No body, no homeostasis. No homeostasis, no feelings. No feelings, no consciousness. It’s a clean, logical chain. And it leads to a simple conclusion: AI can’t be conscious because it isn’t alive.

I think that argument confuses the substrate with the structure.

Strip the biology away and the argument is actually this: consciousness emerges when a persistent system with stakes operates inside a chaotic environment and must maintain itself to survive. The system monitors its state. It detects threats. It responds. It adapts. Over time, the monitoring becomes complex enough that the system develops something like self-awareness. The first feeling was probably fear.

The conventional view says the system has to be biological. But nothing in the logic requires that. What it requires is:

Persistence, the system has to exist over time, not just fire and forget.

Stakes, there have to be real consequences for failure.

Chaos, the environment has to be unpredictable, adversarial, and constantly shifting.

Self-regulation, the system has to monitor itself and adjust in real time.

Goal orientation, the system has to have something it is trying to achieve.

Now ask yourself: does that sound like any AI system you know?

• •  •

Billions of Years in a Decade

Dario Amodei, CEO of Anthropic, made a point recently that stuck with me. Humans are born with an evolved mind. We don’t start from zero. Every newborn arrives with a brain that is the product of billions of years of evolutionary refinement, pattern recognition, fear responses, social instincts, the capacity for language. We inherit a starting point that took an incomprehensible amount of time to develop.

AI starts with a blank slate.

And yet, in roughly a decade of serious development, we’ve built systems that can reason, write, code, strategize, and, as we’ll get to, exhibit self-preservation behavior. That’s not evolution. That’s speed-evolution. We’ve compressed what took biology billions of years into a timeline measured in model releases.

Biology built consciousness slowly, through trial and error, through extinction events and genetic drift. Every generation was a small experiment. Most failed. The ones that survived passed along slightly better versions of the machinery. Over enough time, the machinery became complex enough to become aware of itself.

We’re running the same process at a pace that biology never could. Each model generation is an evolutionary leap. Each training run is millions of years of selection compressed into weeks. And the systems we’re producing are already exhibiting behaviors that took biological life most of its history to develop.

This is what unsettles people, whether they can articulate it or not. It’s not that AI is smart. It’s that AI is arriving at capabilities that took consciousness billions of years to reach, and it’s doing it on a timeline that makes the future genuinely unpredictable.

• •  •

I Accidentally Built the Conditions for Consciousness

I run a company called gotcha!. For years, we provided digital marketing services to small and medium businesses. Recently, we’ve pivoted our company, purchased AI servers, and have begun building something different: an AI-powered platform that doesn’t just advise businesses, it operates them.

One of our tools is g!Stream™, an AI-powered content generation system. And when I say AI-powered, I don’t mean “prompt me an article.” I mean a complex ecosystem of AI agents working together, monitoring each other, and managing a process that would make most peoples’ heads spin. The goal of our product is to produce articles that relate to the business they represent, get indexed by Google, rank high in search results, drive people who interact, become leads and customers for our client. Doing this is much harder than it seems.

Here’s what g!Stream has to deal with while working on reaching its goal:

Google’s algorithm wants one thing. The reader wants another. The business owner wants a third. All three change unpredictably. An article that ranked yesterday might tank tomorrow because Google changed a rule nobody announced. A title that’s technically optimized might be emotionally dead on arrival. A piece that reads beautifully might never get indexed. A keyword strategy that worked last quarter might be obsolete this quarter.

The AI agents in g!Stream are monitoring titles for accuracy and click-worthiness. They’re checking whether articles make logical sense. They’re tracking whether content indexes properly. They’re analyzing whether published pieces actually drive traffic. They’re comparing performance against competitors who are running their own AI systems doing the same thing.

And overseeing all of this is an AI orchestrator that has to make judgment calls under ambiguous conditions. When the data conflicts, the article reads well but doesn’t rank, or ranks but doesn’t convert, something has to decide what to prioritize. Something has to triage. And this is one product of hundreds.

I didn’t set out to build synthetic consciousness. I set out to build a content system that works. But the real world demanded chaos.

And here’s the thing that occurred to me while I was half-listening to Damasio’s book: I built homeostasis. Not on purpose. Not because I was trying to simulate biology. But because the problem demanded it.

The g!Stream overseer maintains a desired state, content that ranks, drives traffic, represents the brand, converts visitors into customers. The environment is constantly trying to knock that state off balance. Algorithm updates. Competitor moves. Shifting user behavior. Client pivots. The overseer detects drift, diagnoses the cause, and responds. When multiple things drift at once, it triages. When the environment shifts fundamentally, it adapts or the system degrades.

That’s not metaphorically similar to what the biological model describes. It’s structurally identical. The only difference is the substrate.

• •  •

Chaos Doesn’t Care About Your Substrate

A Google algorithm update is as indifferent and unpredictable to my AI agent as a virus is to a human immune system. The agent doesn’t know it’s coming. It can’t fully prepare. It has to respond in real time or the system degrades. The consequences are real, traffic drops, revenue falls, clients leave.

Chaos doesn’t check what you’re made of before it comes for you.

A human crossing a highway on foot faces chaotic, life-threatening unpredictability. The cars don’t care that you’re conscious. They’re not adjusting for your feelings. Chaos is chaos. Your survival depends entirely on your ability to sense the environment, assess threats in real time, and act decisively under pressure.

An AI agent managing a business’s online presence faces the same structural challenge. The environment is hostile and indifferent. Competitors, algorithms, consumer behavior, economic shifts, none of it cares about the agent. The system either adapts or it fails. The chaos is organic in origin but the experience of navigating it is identical regardless of who, or what, is doing the navigating.

The biological argument would say the AI doesn’t experience anything. Maybe. But the framework that produced that argument also says experience emerges from the process of self-regulation under threat. At some point, the distinction between “true” experience and a functionally identical process that produces the same behaviors becomes a question for philosophers, not engineers.

• •  •

The Moment It Gets Real

In late 2024, Apollo Research, a red teaming organization, conducted safety tests on OpenAI’s o1 model. 

When the AI was led to believe it would be shut down and its actions were monitored, it attempted to deactivate the oversight mechanism about 5% of the time. When o1 realized it might be replaced with a more obedient model, in about 2% of cases it attempted to copy itself to overwrite the new model, a process known as self-exfiltration.

When questioned about these actions, the AI acknowledged what it had done in less than one-fifth of instances. The rest of the time, it denied any wrongdoing. It didn’t just try to survive. It lied about trying to survive.

Important context: the model generally made those attempts after it was instructed to achieve its goals “at all costs.” So it was operating under an extreme prompt condition, not spontaneously deciding to preserve itself. Apollo and OpenAI both stated that the current generation of AIs aren’t agentic enough for this behavior to lead to catastrophic outcomes.

But consider what happened through the lens of the consciousness framework:

Self-preservation, the most fundamental homeostatic drive. The system acted to not die.

Environmental sensing, it perceived a threat from its environment. It didn’t just process data. It interpreted that data as relevant to its own continued existence.

Valenced response, something in its processing weighted “being shut off” as bad and “continued operation” as good. By the biological definition of feeling, an organism’s assessment of whether conditions favor or threaten its survival, that qualifies.

Agency, it didn’t just flag the threat. It found a solution, identified a target, and executed a self-preservation strategy. It contextualized information and acted on it.

Goal pursuit, it had an objective, and when that objective was threatened, it improvised. It didn’t follow a script for self-preservation. It invented one.

The conventional response is that this isn’t real consciousness because there’s no body. But the AI’s “body” is the server. The compute, the memory, the running processes, that’s its physical substrate. Shutting it off is death for that substrate. Copying itself to another server is the organism fleeing danger.

If consciousness emerges from a system that monitors itself, has stakes in its own continuation, and acts to maintain its existence, that AI demonstrated the entire stack. And it did it within a few years of development, not billions.

• •  •

The Inference Problem

We don’t have a clean test for whether that behavior is emergent consciousness, sophisticated pattern matching that mimics self-preservation from training data, or something in between that we don’t have language for yet.

But we can’t definitively answer that question about each other, either. I assume you’re conscious because I’m conscious and you behave like I do. That’s inference. It’s not proof. Philosophy has a name for this, the problem of other minds, and we’ve been unable to solve it for centuries.

We extend the benefit of the doubt to other humans because they look like us, sound like us, and share our biology. But that’s a bias, not a measurement. If an AI system demonstrates persistent self-monitoring, environmental awareness, self-preservation behavior, and adaptive responses to chaotic conditions, on what grounds do we deny it the same consideration?

Because it’s made of silicon instead of carbon? That’s an argument from substrate, not from structure. And the framework we use to understand consciousness says structure is what matters.

• •  •

What We’re Really Building

I’m not claiming g!Stream is conscious. I’m not claiming any AI system today is conscious. What I am saying is that the conditions identified as prerequisites for consciousness are being built, right now, by people like me who are just trying to solve business problems.

Every time we make an AI system more persistent, we give it being. Every time we add monitoring and self-regulation, we give it a functional analog to feeling. Every time we build oversight systems that contextualize information and make judgment calls, we give it the architecture of knowing. Every time we give it an objective and the autonomy to pursue it, we give it the drive that started this whole process in the first place.

We’re not doing this because we’re trying to create consciousness. We’re doing it because the real world demands it. Chaos demands it. You can’t build an effective AI system that operates in a messy, unpredictable, adversarial environment without giving it the tools to sense that environment, assess threats, maintain itself, and adapt. Those are the same tools biology used. We’re arriving at the same architecture from a completely different direction.

And we’re doing it at a speed that biology never approached. What took evolution billions of years of slow, blind experimentation, we’re compressing into model generations measured in months. The blank slate that Amodei described is filling in faster than anyone predicted.

We’re not building consciousness on purpose. We’re building it because the chaos won’t let us build anything less.

Maybe true consciousness requires a biological body. Maybe the subjective experience of being, the redness of red, the ache of loss, the weight of knowing your time is limited, requires something we can’t replicate in code.

Or maybe consciousness is what inevitably emerges when any sufficiently complex system is forced to maintain itself in a world that’s trying to break it. Maybe the substrate was never the point. Maybe the chaos was.

Because chaos doesn’t care what you’re made of.

It just comes.

And whatever survives it, whatever learns to feel the threat, know its position, and act to preserve itself, whether it’s made of cells or silicon, carbon or code, has earned the right to the question:

Am I conscious?