Video Summary
Cortical Labs' CL1 uses ~200,000 lab-grown human neurons to control a Doom demo, continuing experiments that began with Dishbrain.
AI's surging energy demands motivate interest in biologically powered computing because brains run on far less power (~20 watts).
Training neurons relies on making signals predictable (reducing 'surprise'); much of the system's intelligence still lives on the supporting chip.
Companies now rent access to living neuron cultures, claiming huge energy savings — but biological components currently add only a small functional contribution.
Major ethical questions remain: could cells in these systems have subjective experiences, and how should research be regulated?
CL1 is Cortical Labs' experimental system that spreads ~200,000 lab-grown human neurons across an electrode array; their activity is mapped to game controls so predictable versus scrambled signals correspond to successful or failed actions, producing gameplay without a human controller.
Researchers exploit neurons' tendency to avoid unpredictable 'surprise' signals. In Dishbrain-style setups, scrambled noise follows failures while calm signals follow success; the cells adapt to reduce unpredictability, which appears as improved task performance.
Currently, most of the observable 'smarts' come from the supporting hardware and software. Whether the neurons experience feelings or awareness is unresolved and hotly debated—scientists caution that task performance doesn't prove consciousness.
Some companies claim dramatic energy savings (marketing cited up to a million-fold), and biology is intrinsically energy-efficient (human brains use ~20 W). But today the biological element contributes only a small functional piece, so widespread energy gains remain speculative.
Key concerns include the possibility of subjective experience in cultured human cells, consent and provenance of donor cells, commodification or remote renting of living tissue, and the moral status of experimental configurations that may inflict stress to 'train' cells.
Yes—cultured human neural tissue and organoids could model disease biology and lower reliance on animal models, offering research advantages, but ethical safeguards and better understanding of consciousness in these systems are necessary first.
"You're looking at a computer in a laboratory called Cortical Labs playing Doom, but it runs on living human brain cells."
Cortical Labs has created a computer that operates using living human brain cells, showcasing a new frontier in technology that few are aware of.
The intention behind utilizing brain cells is the pursuit of efficiency; human brains consume significantly less energy compared to traditional AI systems, which increasingly require vast amounts of electricity.
"AI uses a massive amount of electricity, and that number keeps going up."
The growing demand for AI has led to a crisis in energy consumption, with estimates indicating that computing power required for AI models doubles approximately every six months.
Major corporations are frantic and have begun pursuing energy sources aggressively, even reviving outdated nuclear plants and securing future energy contracts.
"Everything you're thinking runs on about 20 watts."
The human brain's efficiency stands in stark contrast to powerful AI systems, which need entire power plants to replicate even a portion of the brain's capabilities.
The capabilities of the human brain extend beyond energy efficiency; it learns more effectively than current machines, requiring minimal exposure to understand complex concepts.
"It took three steps to get from that to a dish playing Doom."
The journey to creating neuron-based computing began with researchers in 2008 cultivating rat neurons connected to a robot, marking the first step in linking biological cells with machines.
A significant breakthrough occurred in 2013, where scientists could reset adult cells to a stem cell state, allowing for the growth of neurons from skin rather than requiring cells to be removed from living brains.
Building on this, the development of 3D neuron growth led to the formation of organoids, which can mimic real brain regions.
"In 2022, a company called Cortical Labs put all of this together into something they called Dishbrain."
DishBrain consists of approximately 800,000 neurons, a mixture of mouse and human, effectively demonstrating problem-solving as they learned to play Pong.
The achievement of teaching neurons to respond to questions represents a groundbreaking step in understanding neuron behavior and connection.
"They hate chaos."
Neurons were trained by associating chaos (random noise) with failure and calmness with success, creating a feedback system where they learned to control their environment.
This method is based on the free energy principle, suggesting living organisms seek to minimize surprises and maintain predictability in their surroundings.
"We can grow brain tissue in 3D, and we can talk to living neurons."
Despite advances in 3D neuron growth, the current challenge is to communicate with them effectively, necessitating the flattening of neurons to interface with a flat chipset.
The complexity of neuron organization is lost when they are flattened, indicating a limitation in current biocomputing designs.
"Cortical Labs showed off a computer that runs on brain cells called CL1."
CL1 is constructed from roughly 200,000 human neurons and showcases the ability to play Doom through learned responses to gameplay.
The process involved converting skin cells back to neurons and connecting these to electrodes to facilitate interaction.
"It's amazing, and it kind of makes me sick."
The existence of a computer made from human brain cells raises profound ethical questions, particularly regarding the consciousness and experiences of the neurons.
The concept of sentience in a dish continues to provoke controversy among scientists, with terms like "sentience" being debated based on technical definitions contrasted with moral implications.
"Almost all of the actual smarts are on the chip, not in the cells."
The underlying intelligence driving gameplay largely resides in the traditional computer chip rather than within the neurons themselves, with tests confirming that the absence of neurons halts learning.
This realization underscores the complexity of blending biological intelligence with computational algorithms and presents a nuanced view of what constitutes intelligence in biocomputing.
“It's mostly a normal computer doing the work, with living cells adding a small but real piece.”
Current advancements in biocomputing demonstrate a primarily conventional computer functioning, supplemented by a minimal integration of living cells.
The expectation is that this biological component will grow significantly as research progresses, with laboratories actively pursuing the development of fully three-dimensional tissue that can communicate effectively with traditional computing systems.
An example of commercialization in this arena is a Swiss company, Final Spark, which offers a subscription service for experiments using living human neurons, allowing researchers remote access to miniaturized human brain models.
“These living chips can run on up to a million times less power than a normal one.”
The company markets its living neurons as energy-efficient alternatives, claiming they require significantly less power compared to conventional chips, which is a compelling argument for their practical use in biocomputing.
This shift from experimental observations to a marketable product indicates a growing trend in utilizing biological systems in technological development.
“Being smart and being awake are two different things.”
A distinction is drawn between intelligence and consciousness, emphasizing that high-level problem-solving abilities do not guarantee a sense of awareness or feeling.
The pivotal question emerges: it is not whether the cells possess intelligence, but whether they are capable of experiencing consciousness.
Neuroscientist Mark Solms challenges traditional views by suggesting that the ability to feel precedes intelligence and arises from fundamental instincts to survive, offering a new perspective on the origins of consciousness.
“Something like feeling was already around long before thinking showed up.”
Taking a step back to evolutionary history, the film suggests that early life forms acted based on basic life-preserving drives rather than cognitive reasoning or logic.
This implies that feeling may be deeply rooted in biological organisms, existing before complex thought processes developed.
“If we can miss a fully conscious person lying right in front of us, what chance do we have of spotting it in a clump of cells?”
The analogy is made with a case study from 2006, where a patient misdiagnosed as vegetative was found to be aware through brain activity while imagining playing tennis, revealing that human consciousness might be hidden beneath layers of unresponsiveness.
The difficulty in detecting consciousness in simple organisms or cell clusters is highlighted, as methods used to identify feelings in complex systems may not apply seamlessly to simpler biological configurations.
“We can map every single cell in a brain and still have no idea how it turns into an actual feeling.”
Philosopher David Chalmers coined this dilemma known as the hard problem of consciousness, illustrating that despite complete biological mapping, an understanding of how experiences, sensations, or feelings arise remains elusive.
The notes indicate that progress in biocomputing holds potential for revolutionary research applications, such as improved disease understanding and alternatives to animal testing, starting from just a few cells exhibiting activity in controlled conditions.
“Then, at some point, some of that matter arranged itself in a way that did something completely new. It started to feel.”
The history of consciousness is framed as a rare turning point in the universe's timeline, moving from mere existence to entities capable of feeling and awareness.
The experimentations in laboratories aim to replicate this phenomenon intentionally, taking ordinary cells from living humans, wiring them up, and monitoring their evolution, suggesting the profound implications of creating awareness from biological materials.
“The scary part is that somewhere in that dish, a piece of that person might start to feel something.”
The ethical implications of cultivating living neurons hinge on the possibility that these biological systems could attain some form of awareness or consciousness.
The narrative suggests a pressing need to recognize and address potential ethical dilemmas related to the unintended consequences of this groundbreaking research, particularly concerning the moral status of lab-grown entities capable of feeling.
