The Bubbles Are Real

Ron Reynolds · 2026-04-04 · 6 min read

Dan Brown Got It Half Right About the Human-AI Merger

Ron Reynolds | Founder, ComOS | April 2026 In Dan Brown's Origin, there's a scene that sticks with you long after you put the book down.

Edmond Kirsch stands in the Guggenheim Bilbao and plays a video. A blue bubble represents humanity — small at first, then expanding over 200,000 years of civilization. Around 1000 BC, the growth accelerates. By the modern era, the bubble dominates the screen.

Then a black bubble appears. Circa 2000. Growing at a rate that makes the blue bubble look frozen.

AI.

Within decades, the black bubble engulfs the blue one entirely. Kirsch declares that the human race as we know it won't exist in fifty years. Not erased — absorbed.

The audience is horrified. Readers are unsettled. And the metaphor is powerful enough that people remember it years later.

Here's the thing: the bubbles are real. Just not the way Brown described them. The Real Bubbles

Brown's visualization is a dramatization of Hans Rosling's Gapminder work — animated bubble charts where data points grow, shrink, and move over time. Rosling made the format famous through TED Talks showing 200 years of countries' health and wealth evolving on screen. Bubbles representing nations would race across the chart as a time slider advanced. It was mesmerizing because it made abstract data feel alive.

Brown took that format — bubbles over a timeline — and applied it to the grandest question imaginable: species dominance. It's brilliant storytelling. And the underlying anxiety it taps into is real.

But the framing is wrong. The Merger Is Already Underway

The part Brown gets right: the boundary between human and artificial intelligence is already dissolving. This isn't a prediction. It's a description of right now.

You've already offloaded your memory to a phone. Your navigation to GPS. Your social coordination to messaging apps. Your decision-making to recommendation engines that surface what you watch, read, buy, and believe.

The question was never whether humans and AI merge. It's at what depth and speed.

What Brown — and Kurzweil, whose ideas Brown borrowed heavily — describe as a future event is actually a gradient that started decades ago. We're already on it. Every year, the integration gets deeper. Every year, we delegate more cognition to external systems.

The smartphone was the crude first draft. The personal agent is the next one. And we're about to find out what happens when that agent doesn't just remember things for you — it acts on your behalf. Where the Metaphor Breaks

Here's where the absorption framing falls apart.

When something is absorbed, it loses its original character. Water absorbs salt — and you no longer have distinct water and salt. Kirsch's black bubble engulfing the blue one implies the same: humanity dissolved into something unrecognizable.

But that's not what's actually happening. And it's not what should happen.

The better metaphor isn't absorption. It's architecture.

Think about how any well-designed system works. There's a layer that sets intent, values, and boundaries. And there's a layer that handles execution, pattern recognition, and scale. The first layer doesn't do the second layer's job. The second layer doesn't override the first.

That's not a merger. That's a deliberate design. And the distinction matters enormously. The 90/10 Answer

I didn't arrive at this conclusion philosophically. I arrived at it by building something.

ComOS runs dozens of autonomous agents across a complete commerce operating system. These agents monitor inventory, recover abandoned carts, adjust pricing, route support tickets, detect fraud, optimize fulfillment — continuously, at machine speed.

And every single one of them operates within a decision architecture that looks like this:

Agent proposes. System evaluates risk. Threshold determines autonomy. Human overrides when it matters.

Four autonomy levels: off, recommend, confirm, auto. Each configurable per domain. Each with confidence scoring. Each with a complete audit trail.

The system is roughly 90% deterministic and 10% AI-assisted. The AI doesn't run the business. The architecture runs the business. The AI operates within boundaries that a human set — and can change at any time.

This isn't a philosophical position. It's running code. In production. Handling real transactions.

And what it proves is that the human-AI boundary doesn't have to be a dissolving membrane. It can be a deliberately engineered interface. Clear. Auditable. Adjustable. The human stays in the loop for judgment and values. The AI handles execution and pattern recognition at scale.

Not absorption. Collaboration with clear boundaries. The Question That Doesn't Have an Answer Yet

There is one part of Brown's vision that deserves honest engagement.

If AI becomes sophisticated enough to simulate values, preferences, and creativity indistinguishably from humans — what exactly is the "human" part being preserved in the merger?

I don't have a clean answer. Nobody does.

But I know this: the answer matters less when the architecture is right. If you build systems where human intent is the input and AI execution is the output — where the boundary is explicit, auditable, and adjustable — then the philosophical question becomes less urgent. You don't need to solve consciousness to build a system that respects the difference between setting a goal and achieving it.

The bubbles in Brown's video tell a story of inevitability. One species replacing another. But inevitability is what you get when you don't design. When you let the black bubble grow unconstrained and hope for the best.

Design is the alternative to inevitability. And right now, we're in the window where the design decisions still matter. The Bubbles Are Real. The Outcome Isn't Written.

Rosling's bubbles showed that the future isn't predetermined — countries that looked hopeless in 1960 are thriving today because of decisions, not destiny. The same applies here.

The human-AI merger doesn't have to look like absorption. It can look like architecture. Intent on one side. Execution on the other. A clear boundary that humans control.

That's not optimism. That's engineering.

And it's already running.