Wetware Computing: Human Brain Cells Learn to Play Doom

While the biological system performs far below the level of experienced human players, the experiment highlights the rapid progress of Wetware computing, a new field where living neural tissue and electronic hardware work together to process information.

What Is Wetware Computing?

Wetware refers to computer systems that use living biological neurons as part of the processing architecture. Instead of relying entirely on silicon chips and transistors, these systems cultivate brain cells in laboratories and connect them to electronic interfaces.

The neurons grow across microelectrode arrays that allow computers to send electrical signals to the cells and record the activity that comes back. As the network responds to stimulation, the neurons begin forming connections and adapting their behaviour, effectively learning from feedback in ways that traditional computing hardware cannot easily replicate.

This hybrid approach is sometimes described as Synthetic Biological Intelligence (SBI), where biological networks and digital software operate together as a new type of computing system.

From Pong to Doom

The recent Doom experiment builds on earlier work by Australian biotechnology company Cortical Labs. Several years ago, the team created a neural computing system where roughly 800,000 neurons grown in culture were trained to play the classic arcade game Pong.

Those early demonstrations required years of laboratory experimentation and highly specialised equipment. The neurons had to be trained to control paddles on a screen through carefully designed patterns of electrical feedback.

The latest breakthrough represents a significant leap forward. With new interfaces that allow developers to interact with the system using the Python programming language, an independent developer was able to teach the neuron network to interact with Doom in roughly a week.

Doom is far more complex than Pong. It requires navigating environments, reacting to moving objects and making real time decisions, making it an important test for biological computing systems.

How the Neurons Learn

The learning process relies on the natural behaviour of neurons seeking predictable patterns.

When the neural network produces responses that lead to stable or predictable outcomes, the system reinforces those connections. When signals become chaotic or random, the network gradually adapts to avoid those patterns. This feedback driven learning mechanism encourages the neurons to organise themselves into pathways that perform useful tasks.

Although the biological system is still primitive compared with modern gaming AI, researchers have observed that these living networks can learn surprisingly quickly, sometimes adapting faster than traditional machine learning systems built purely on silicon processors.

The CL1 Biological Computer

The Doom experiment is possible because of advances in hardware platforms designed specifically for Wetware computing.

Cortical Labs recently launched the CL1 biological computer, a commercial system that integrates living neurons with electronic computing hardware. The platform allows researchers to grow human neural networks directly on silicon chips and interact with them through software.

Inside the CL1 unit, neurons are placed on a planar electrode array that acts as both input and output for electrical signals. A life support system maintains the cells using temperature control, nutrient circulation, filtration and gas mixing to keep the culture stable.

The system can even run without an external computer, and multiple units can be combined into server racks containing dozens of biological neural networks operating simultaneously.

Researchers can either purchase the hardware directly or access it remotely through a cloud based platform, a model the company describes as Wetware as a Service, allowing scientists around the world to experiment with living neural systems without building their own laboratories.

How the Neurons Are Created

The neurons used in these systems typically originate from induced pluripotent stem cells, often derived from blood samples. These stem cells can be guided to develop into neurons using specialised differentiation techniques.

Scientists can either use molecular signals that mimic the conditions of a developing brain or directly activate specific genes associated with neuron formation. Each method produces slightly different cell populations, which researchers are still studying to determine which combinations produce the most capable neural networks.

Understanding these differences is central to one of the field’s biggest research questions: what scientists call the Minimal Viable Brain. This concept explores the smallest possible network of neurons capable of producing intelligent behaviour.

Some organisms demonstrate surprisingly capable behaviour with only a few hundred neurons. The tiny worm Caenorhabditis elegans, for example, operates with just over 300 neurons. Researchers are exploring whether artificial biological networks could achieve similar complexity with far fewer cells.

Why Wetware Matters

Although playing Doom may appear trivial, it demonstrates that biological neural networks can cope with complex environments, uncertainty and real time decision making.

Those capabilities are critical for a range of potential applications, including:

• controlling robotic limbs or prosthetics
• studying neurological diseases such as epilepsy or Alzheimer’s
• accelerating drug discovery and testing
• building energy efficient forms of artificial intelligence

Biological neural systems may also help scientists understand how intelligence itself emerges from networks of cells.

Ethical and Scientific Questions

The rise of Wetware computing also raises new ethical questions. Because these systems use living human derived cells, researchers must carefully consider issues around consciousness, sentience and responsible experimentation.

For now, the neural cultures used in these experiments are extremely simple and far removed from functioning brains. Still, research groups have already begun developing ethical frameworks and safeguards to guide how these technologies are used.

The Future of Wetware

Wetware computing is still in its early stages, but progress is accelerating quickly. The development of programmable platforms such as the CL1 system means researchers can now experiment with biological computing in ways that were impossible just a few years ago.

If these systems continue to evolve, the next generation of computers may not rely solely on silicon chips. Instead, they could combine digital electronics with living neural networks that learn, adapt and process information in ways traditional machines cannot.

The experiment that taught brain cells to play Doom may one day be remembered as an early glimpse into that future.

The biggest risk right now is doing nothing, but the second biggest risk is doing it recklessly. If your competitors are using agents to handle lead gen while you’re still manually typing into a chat box, you will lose ground. However you want to work cautiously and not rush into using AI agents without considering the security aspects.

If you’re feeling overwhelmed by names like Claude, OpenClaw, and RCE vulnerabilities, that’s where we come in. Our AI Workshops and consulting take the hype and the danger out of the equation. We show you how to build a digital workforce that is as secure as it is productive. Ready to get started? Get in touch with our team today!