AI Isn’t a Brain but Soon It Could Be

In 1943, research scientists Warren McCulloch and Walter Pitts described neurons as binary switches, either “on” or “off,” laying down the theory for artificial neural networks. Though their goal was to better understand the brain, their model set the stage for artificial intelligence (AI). And as we all know, the AI industry is now producing tools that simulate speech, predict protein structures, and generate art. But today’s AI is not much like a brain. It is based on math, statistics, and pattern recognition. It remains fundamentally a different way of processing compared to the human brain.

Current AI spots patterns in data, but it’s not curious, emotional, or able to explore. Your brain is different; it is a living, evolving biological system, not just code. It uses chemicals and hormones, not just electricity, to adapt and respond to the world. Neither ChatGPT nor any other current AI model can do that.

Then There Was Fire

Geoffrey Hinton, a key figure behind the development of AI, has led the research in neural networks. Some even say his impact could surpass the discovery of fire, which sounds wild, but has some truth to it, as AI begins to reshape everything.

However, primitive nervous systems, like those of slugs, display adaptive intelligence that AI cannot match. A maggot’s neuron, for instance, receives input from experience and adjusts its behavior in nuanced, analog — not digital — ways. AI’s digital neurons are mathematical abstractions that form functions based on uploaded data. While useful, this processing lacks the flexibility and responsiveness of biology. At its core, AI remains a statistical engine, not a living, thinking organism with desires or curiosity.

Enter the ‘’Free Energy’’ Principle

Hinton and Karl Friston were both at Cambridge and worked just down the hall from each other. They were both exploring the neuroscience of intelligence from different angles. Hinton developed deep learning and backpropagation for neural networks. Friston developed the Free Energy Principle, which says that the brain learns by reducing surprise. In 2010, this work was published in the journal Nature, proposing a unified brain theory. Friston's model of “Free energy” explains the gap between what the brain expects and what actually happens. This shows how the brain constantly updates its understanding of the world using a process called "active inference." Unlike Hinton’s AI model, which analyzes data, Friston shows how the brain predicts and learns through real-world experience. Friston was laying the foundation for a very different kind of artificial intelligence, one based on "natural intelligence" of how the brain works.

The Emergence of Natural Intelligence

AI typically processes past data to make predictions. It does not explore new ideas. This absence of real-time feedback and updating is a major distinction between human cognition and machine learning. In ChatGPT, the GPT stands for Generative Pre-trained Transformer. In other words, AI is pre-trained on large data sets. As AI generates text, it will predict the most likely next output based on patterns it spots in the data it was given.

But that is changing.

AI based on Friston’s modeled on the Free Energy Principle has been described as using "Natural Intelligence." His theory is now a new platform that does not merely predict based on previous data; it actually reasons, plans, acts, and learns continuously through a real-time intelligence loop.

Here is how it works: Traditional AI platforms like ChatGPT, Claude, or Gemini make predictions based on patterns found in the data they have been trained on. They process vast datasets to generate statistically likely outputs. But probability is not the same as understanding. That is why these systems sometimes hallucinate when faced with unfamiliar input, and so they may produce a confident but incorrect response.

The new approach, guided by Friston’s work, is very different. It behaves more like a curious scientist or a child encountering something for the first time. When faced with uncertainty, it does not just guess; it investigates. In contrast, current AI models simply connect patterns they recognize in past data.

For example, when it hears "ruff-ruff," this new kind of AI considers multiple possibilities: a dog, a fox, or even a human imitating a canine. Unlike statistical models, it seeks additional input: Can it see fur? How does the dog respond to a whistle or seeing a cat? This AI updates its understanding with each interaction. And that is exactly how the brain learns, through trial and error.

Artificial Intelligence Essential Reads

Do You Have Bixonimania?

The Skill AI Can’t Replace: Agency

The Intelligence Loop and Practical Applications

This approach is enabled by the intelligence loop: a four-step cycle of understanding the world, reasoning about it, planning actions, and learning from outcomes. The loop repeats endlessly, allowing the platform to refine its knowledge continually. Unlike generalist current models, it adapts to the needs of the user. Current LLMs (Large Language Models) go broad but shallow. This new approach goes deep and narrow. It becomes an expert. It models cause and effect, and reasons forward, not just looking backward at the data it was fed.

This is what makes the Free Energy Principle work. In logistics, it adapts routes in real time based on current traffic, weather, and even unexpected road closures, not just past data. In healthcare, it personalizes treatment by observing how each person’s body responds to medication, adjusting in real time rather than relying on population averages. In retail, it can tailor inventory and staffing decisions dynamically by learning from live customer behavior, not just historical sales trends. It learns what works for you, in real time.

This approach shifts AI from simply simulating human thinking to actually making intelligent, real-time choices. While traditional AI mimics, Friston-inspired systems understand. This new model can explore, adapt, and has an opinion based on what is experiences.

Friston’s vision brings neuroscience into the heart of AI. It does not replace current models; it complements them. As AI continues to evolve, it may be the principles of living brains, not just bigger datasets, that truly move us forward. As we apply the Free Energy Principle, May the Force be with You.