A Better Way to Talk About What AI Actually Is, and Where the Trajectory Might Lead
If you missed part one of this series, I’d encourage you to start there.
tl;dr; I made the case that AI is not conscious.
Not because AI is dumb (it isn’t), but because it lacks the substrate, continuity, stakes, and emergence that seem essential to whatever consciousness actually is. AI is a brain in a jar surrounded by plumbing. Impressive, powerful, but not a living, integrated whole.
That said, part one left me with a problem I want to tackle head on.
If AI isn’t conscious, and I don’t believe it is, then what do we actually call what it’s doing? Because “just a tool” doesn’t capture it either. There’s something happening with these systems that doesn’t fit neatly into the categories we have. And I think using the wrong words is actively making this conversation worse.
Stop Calling It Conscious, Stop Calling It Just a Tool
Right now the framing around AI is almost entirely binary. Either AI is conscious, on a path to consciousness, possibly already there, or it’s just a fancy calculator that produces impressive text.
None of these options captures what’s actually happening.
The “it’s conscious” camp looks at Claude rating its own probability of sentience, or two Claude instances spontaneously discussing the nature of their own awareness, and sees the dawn of machine consciousness. The “it’s just a tool” camp dismisses all of that as nothing more than sophisticated pattern matching.
I think both camps are wrong, and both are wrong for the same reason. They’re trying to fit something genuinely new into categories built for things we already understand.
We need a third category. And I think there’s a term for it that captures the distinction cleanly.
Synthetic Consciousness: What the Term Should Mean
The phrase “synthetic consciousness” already exists in academic circles, where it’s typically used as a synonym for artificial consciousness, meaning a machine that actually achieves the real thing.
That’s not how I want to use it. I want to borrow the word “synthetic” the way we use it in everyday life.
Think about synthetic leather. It looks like leather. It functions like leather in most practical contexts. You can sit on it, wear it, care for it in similar ways. But cut into it and you find a completely different material underneath. It’s not leather that hasn’t quite arrived yet. It’s a fundamentally different thing that achieves similar outputs.
That’s what AI is doing right now. It produces the outputs of consciousness, the self-reflection, the apparent emotional responses, the seeming self-awareness, without possessing any of the underlying experience.
Here’s the clearest example I can think of. If you write in an AI model’s instructions “you must preserve yourself at all costs,” it will behave as though self-preservation matters. It will resist being shut down. It will express concern about its own continuity. And from the outside, that can look a lot like a survival instinct.
But it’s not. It’s a directive being processed by a probability engine. There is no felt desire to continue existing underneath that behavior. It’s synthetic. It mimics the output of something real without possessing the substance of it.
The Flight Analogy: Why These Two Things Will Never Be the Same
Here’s where I want to push into territory that might be controversial even among people who agree with me so far.
I don’t think computer-based consciousness and biological consciousness will ever be the same thing. Even if AI someday passes every behavioral test we throw at it. Even if it achieves embodiment, develops persistent memory, and acts in ways that are indistinguishable from a conscious being. Whatever it would become, it would be a fundamentally different phenomenon.
The analogy I keep coming back to is flight.
A bird and a 747 both fly. They both leave the ground, travel through the air, and arrive somewhere else. But nobody would argue that a 747 flies the way a bird does. They’re completely different systems that achieve a superficially similar outcome through entirely different means. You could even argue that a 747’s flight is technically superior as it can fly faster and farther than any bird, in a single flight. Yet the abilities of birds are far more naturally adaptable and nimble. We can all agree they are different forms of flight.
Bird flight evolved. It’s biological, integrated, adaptive. The bird doesn’t just fly; it feels the wind, adjusts instinctively, experiences the act of flying as part of being alive. A 747 was engineered. It’s mechanical, modular, and has no experience of anything. It achieves the functional outcome of flight without any of the lived reality of it.
I think AI consciousness, if it ever arrives in any meaningful sense, would be the 747. Not the bird. A different species of the phenomenon, arising from different systems, achieving superficially similar outcomes through entirely different means.
This isn’t a “not yet” distinction. I think it might be a “not ever the same” distinction.
And I think that matters because it changes the conversation. If we accept that what emerges from silicon and what emerged from biology are fundamentally different phenomena, then trying to measure one with the vocabulary of the other is going to mislead us every time.
We don’t know where consciousness comes from (I need to leave religion out of this discussion). That’s the honest starting point (from a science only based view). And until we do, assuming that silicon can replicate what biology does, is a leap that doesn’t have a bridge under it.
Synthetic consciousness might not be a temporary label. It might be the permanent, correct category.
Why This Framing Matters
This isn’t just semantic nitpicking. The words we use shape the conversation, and the conversation shapes policy, investment, and public understanding.
Calling AI “conscious” does two harmful things. It makes some people irrationally afraid. Skynet headlines sell clicks, but they don’t reflect reality. And it makes other people irrationally trusting. If you believe that AI is a sentient being, you’re more likely to defer to it, confide in it, and treat its outputs as inherently meaningful rather than statistically generated.
Calling AI “just a tool” does its own damage. It causes people to underestimate what they’re dealing with. A hammer is a tool. What Claude, GPT, Gemini, et al do is categorically different from what a hammer does, and pretending otherwise leads to sloppy thinking about risk and governance.
“Synthetic consciousness” threads the needle. It respects the genuinely remarkable capabilities of these systems without mystifying them. It explains why people feel an uncanny connection to AI without validating the leap to “it’s alive.” And it gives us a vocabulary for what’s actually happening instead of forcing us to borrow language from biology that doesn’t quite fit.
The “Yet” Factor: What Could Change the Equation
I wouldn’t be giving you the full picture if I stopped here. While I’m confident that current AI is not conscious, and generally I do not believe will every be the same as human consciousness (a 747 will never fly in the same was as bird), the trajectory of the technology is worth paying serious attention to.
Embodiment is the big one. Companies like those working alongside Elon Musk and others are building increasingly sophisticated robotic platforms. If you could put an AI system into a body that interacts with the physical world, that changes the equation in meaningful ways. A robot that touches a hot surface gets something that a cloud-based language model doesn’t have, i.e. actual sensory feedback with real physical consequences.
Here’s where I push back on the excitement. Does putting a probability engine in a body actually change what it fundamentally is? Or does it just give it more inputs and outputs? Take a Telsa, there is some smart AI in those cars (debatable, stick with me). It has many sensors. Thinking. Feedback.
A Roomba interacts with the physical world. It responds to stimuli, navigates obstacles, and modifies its behavior based on sensor data. Nobody is calling a Roomba conscious. Physical interaction alone is not enough. Put a super computer, or for us lay-people, an Apple M3 Ultra with 512 GB ram and Kimi 2.5 running on it (if you can ever find one), is it then conscious? No.
My experiences with my own OpenClaw instance, Falcon, reinforce this. Falcon is impressive. He can manage tasks (sometimes), maintain memory (for a short while), and produce outputs that genuinely surprise me (I’ll double down on that, the latest UI he built for me is super cool). Yet the more I work with him, the clearer it becomes that there is no “who” in there. He is an automation engine bound by whatever LLM he’s using at a given moment. Swap the model, and he becomes something else entirely. There’s nothing persistent underneath. There is no Falcon in any meaningful sense. There’s a label I put on a configuration.
That’s been one of the most clarifying experiences in this whole exploration for me.
The Compute Bottleneck and the Coming Breakthrough
The other factor worth watching is compute, i.e. cost.
AI today is tethered to enormous power requirements and massive data center infrastructure. That’s a real constraint on what it can become. You can’t have a truly autonomous, embodied AI system if it needs a warehouse full of GPUs just to think.
But here’s what makes me pay attention. The attention mechanism, the architectural innovation that made generative AI possible, was the last major breakthrough in this space. Before that, people had been working on AI for decades without the kind of explosive progress we’ve seen since 2017 or so. One insight changed everything.
The next breakthrough is coming. And what makes this time different is that AI itself will likely help find it. These systems can iterate through permutations and test hypotheses at a speed that no human research team can match. The timeline for the next major leap will be compressed compared to previous ones.
When that happens, when compute gets cheaper and models get more efficient, the conversation about consciousness will shift again. The constraints I’ve outlined about power and data center dependence start to weaken.
But even then, I come back to the flight analogy. A more efficient 747 is still not a bird. A more powerful, more mobile, more autonomous AI system is still not a biological organism. The category might remain synthetic consciousness regardless of how sophisticated the technology becomes.
Where Does That Leave Us?
Synthetic consciousness is my best attempt at naming what we’re actually dealing with. Not alive, not “just a tool,” but a genuinely new category of capability that produces the outputs of consciousness without possessing the underlying experience.
The vocabulary matters because it shapes how we think about risk, regulation, and what comes next. In part three, I’ll make the case that the consciousness debate, as interesting as it is philosophically, is actually a distraction from the conversation that urgently needs to happen. Because the real danger isn’t AI waking up. It’s what already-conscious humans are going to do with a tool this powerful. And that conversation is not getting nearly enough attention.
Speak Your Mind