Intelligence.

What do we actually mean when we call something intelligent?

Not in the everyday sense of calling someone intelligent because they are good at mathematics, have a broad vocabulary, or know a great deal about a certain subject. I mean intelligence itself. What is the underlying thing we are actually describing?

Starting with a cell

To explore that, I want to start as simply as possible: with a single-celled organism.

It has DNA, and my first instinct is to think of that DNA as a kind of model. It contains inherited information about how the organism should function, survive, respond to certain conditions, and reproduce. That is probably not quite precise enough, though. DNA itself is not really a model of the world in the same way a brain might construct one. It is closer to a compressed rulebook, shaped by evolution over countless generations.

Still, when you look at the organism as a whole, something interesting happens. Through receptors, chemical feedback loops, gene expression, and changes in its internal state, it can detect changes in its environment and respond to them. It can move toward favorable conditions, away from harmful ones, regulate itself, and alter its behavior depending on what it encounters.

So the first question appears almost immediately: is that intelligence?

If intelligence requires a complete understanding of the world, then obviously not. But that definition creates a problem, because humans do not understand the world in its entirety either. We understand far more than a single-celled organism, but our own understanding is still partial. We build models, test them, improve them, and use them to survive and act within the part of reality we can access.

We do not understand the universe as a whole. I would argue that we do not even know, in any final sense, what the universe is. In that respect, perhaps the gap between us and the single-celled organism is not as absolute as it first appears.

Intelligence in layers

A human being is, after all, itself a collection of cells. None of those cells understands the organism it belongs to. A neuron does not understand language. A muscle cell does not understand why the body is walking across a room. Yet together, these cells form something with abilities that simply do not exist at the level of the individual cell.

That makes me wonder whether intelligence can emerge in layers.

The same pattern seems to appear when we move from the individual human to humanity as a whole. No single person understands modern civilization in its entirety. No one person could recreate our medicine, engineering, agriculture, mathematics, computer science, legal systems, or accumulated history from scratch. Yet all of that knowledge exists within humanity.

What makes this possible is our ability to share information.

Language allowed knowledge to move between people. Writing allowed it to survive the death of the person who first learned it. Books expanded that shared memory even further. Printing made knowledge easier to distribute, and later technologies such as telecommunications, film, computers, and the internet made that exchange faster and broader than ever before.

The result is something no individual human could ever become on their own. Humanity can preserve discoveries for thousands of years, compare ideas produced on opposite sides of the planet, correct old mistakes, and build new knowledge on foundations laid by people long dead. A scientist today does not begin from nothing. Neither does an engineer, a doctor, or a historian. Each of them enters a system that already remembers.

In that sense, civilization starts to resemble a larger information-processing organism. The comparison is imperfect, of course. Humanity is nowhere near as coordinated as a body, and our individual interests frequently conflict. But the general pattern remains: limited units become connected, information is exchanged, memory is preserved, and capabilities emerge that no individual unit possesses.

Perhaps that means intelligence is not something that exists only inside a single mind.

A cell responds to its environment. A multicellular organism combines billions of cells into something more capable. A human builds abstract models of the world. Humanity then connects billions of those minds and preserves their discoveries across generations.

Seen that way, civilization itself begins to look like another layer of intelligence.

From memory to reasoning

Artificial intelligence becomes interesting when placed inside that same sequence.

At first glance, AI can be understood as just another tool for moving information around. Writing gave us external memory. Libraries gave us somewhere to store it. The internet made that memory accessible almost instantly from anywhere in the world. Artificial intelligence could simply be the next improvement in that process: a faster and more convenient way to retrieve information.

But I am not sure that description goes far enough.

A book stores information, but it cannot combine that information in response to a new question. A search engine can find relevant sources, but it does not necessarily build an answer from them. Artificial systems are beginning to do something different. They can compare ideas, identify patterns, summarize large amounts of information, reason across different domains, and produce answers to questions they have never encountered in exactly the same form before.

That starts to look less like storage and more like processing.

If civilization can be thought of as a larger information-processing system, then AI may represent something more significant than another communication technology. It may be a new cognitive component inside that system.

The distinction matters. Humans have spent thousands of years creating ways to preserve and share thought, but the thinking itself still had to happen inside a human mind. With AI, part of that process can now happen outside us. We can hand a problem to an artificial system, ask it to compare possibilities or construct a model, and receive something that resembles a piece of reasoning in return.

Between a tool and an agent

That raises another question. Is the artificial system itself intelligent, or is it merely extending human intelligence?

I suspect the answer may be both.

An AI system can show properties that I would already be willing to call intelligent. It can generalize from what it has learned, respond to unfamiliar situations, discover patterns, and solve certain problems without being explicitly given every step in advance. If those same abilities appeared in an animal, we would probably have little hesitation in describing them as signs of intelligence.

At the same time, I think the word intelligence is often used too casually when we talk about current AI systems. The name “artificial intelligence” makes it tempting to imagine that we have already created something equivalent to an independent intelligent being. That is not really what most current systems are. They can display intelligent behaviour without possessing the continuity, autonomy, experience, or persistent goals that we normally associate with an intelligent agent.

Perhaps the problem is not that calling these systems intelligent is entirely wrong, but that the word is doing too much work. We use the same term for the ability to solve a problem, for the broader capacity of an organism to adapt to its environment, and for the idea of an autonomous mind. Those are not necessarily the same thing.

Today’s artificial systems still depend heavily on the larger human system around them. Their knowledge comes largely from information produced by people. Their goals are usually given to them by people. Their opportunities to act in the world are limited by the tools and permissions we provide.

So perhaps it makes more sense to see current AI as something between a tool and an independent agent.

It is not merely a passive store of information, but neither is it quite an organism moving through the world under its own direction. It occupies a strange middle ground: capable of reasoning, but usually waiting to be asked; capable of building models, but without a continuous life in which those models are constantly tested against the world.

Continuity and experience

For that to change, an artificial system would need something closer to continuity. It would need persistent memory, an ongoing model of its environment, the ability to act, and the ability to observe what happened as a result. It would need to learn not only from a training process conducted beforehand, but from its own continuing experience.

Most importantly, it would need goals that persist over time.

At that point, the comparison with living intelligence would become much harder to dismiss. Such a system would perceive, remember, predict, act, experience consequences, and adjust. In functional terms, that would begin to look very much like the process we have been calling intelligence all along.

Which leaves one particularly difficult question untouched.

Would any of that mean it was conscious?