From: mk_thisisit
Biological processors, made from living cells, represent a groundbreaking advancement, with no historical precedent in their creation [00:00:01]. These processors are grown on special plates to form three-dimensional structures [00:00:15]. This method is the most energy-efficient known [00:00:23], being about a million times more efficient than traditional methods [00:00:28].
Final Spark: Pioneering Biocomputing
Final Spark, a Swiss biocomputing startup founded in 2014, is the first in the world to build a living processor from living neurons [00:01:05]. The company aims to combine the computing power of traditional computers with the efficiency of the human brain [00:01:16].
Human Brain Organoids
A human brain organoid is a small fragment of nervous tissue composed of neurons derived from humans [00:01:26]. While many institutions worldwide work on human brain organoids, Final Spark’s unique contribution is their application for computing [00:02:26]. The core assumption is that human neurons can be used for computation [00:02:49].
The human cells used are pluripotent stem cells, typically derived from fibroblasts (skin cells) that are reprogrammed to a state of pluripotency [00:03:10]. This allows them to become any cell type [00:03:22]. Specific molecules are added to the culture medium to differentiate these pluripotent stem cells into neurons, astrocytes, and other brain cells [00:03:36].
Organoids are grown in an incubator under precise conditions of temperature and carbon dioxide concentration [00:04:39]. They are supplied with a special culture medium containing nutrients [00:05:51] and maintained in a flow system with a pump circulating the medium [00:06:23]. The internal temperature is kept at 37 degrees Celsius, like the human body [00:06:53]. Final Spark has successfully kept these cells alive for almost three years [00:07:06].
Connecting Organoids to Neural Platforms
Organoids are placed directly onto a multi-electrode array (MEA) to record their activity [00:07:17]. A membrane with a hole, called “confetti,” helps position the organoid precisely and reduces system noise [00:08:39].
Driving Force: Energy Efficiency for AI
The decision to pursue biocomputing stemmed from the energy consumption of artificial neural networks [00:09:34]. Simulations of 100 neurons in traditional AI systems can consume several kilowatts of power [00:10:01]. In stark contrast, the human brain, with 86 billion neurons, requires only 20 watts [00:10:07]. This massive disparity highlighted the need for a more sustainable approach to AI development [00:10:25]. The main motivation was fundamental research on artificial intelligence, realizing that using living neurons was the best way to advance it [00:10:31].
Demonstrating Biocomputing Capabilities: The Butterfly Experiment
The “butterfly experiment” allows browser-based users to interact directly and in real-time with brain organoids [00:11:16]. Users send stimulation to the organoids, which is visually represented by a blue dot [00:11:38]. A butterfly on screen moves in three dimensions, and its flight direction is controlled by the organoid’s response [00:11:48]. If the butterfly responds to stimulation, it flies straight; if not, it moves randomly [00:12:10]. This experiment demonstrates the communication established with the organoids, showing bursts of activity where multiple neurons transmit information and process it [00:12:47].
To induce information processing and enforce specific behaviors (learning), Final Spark uses dopamine to reward a brain organoid when it performs a desired action [00:13:03].
Future Implications and Challenges
Biocomputing is seen as the future of data processing, particularly for artificial intelligence calculations [00:17:09]. Similar to how quantum computers have specific applications, biocomputing can be used for particular types of computations, with AI fitting perfectly due to its neural network basis [00:17:13]. This technology has the potential to transform AI development and drastically reduce its carbon footprint [00:17:31].
Ethical Considerations
Final Spark acknowledges the ethical issues surrounding their work [00:44:00]. While the scientists focus on technical questions, they engage with universities and ethics experts to consider the implications of using living neurons for computational purposes [00:14:52].
Why Human Cells?
While rat cells could potentially work for computations, the use of human cells opens up the possibility for future therapeutic applications [00:15:39]. The company prefers that any discoveries made along the way could help people rather than just rats [00:16:04].
Key milestones for the company include successfully recording neuronal activity, growing cells from liquid nitrogen, and observing neural tissue responses to dopamine [00:16:21]. The most challenging aspect has been simply keeping the cells alive, which was initially only possible for a few hours but now for weeks [00:14:05]. The CEO of Final Spark, Fred, previously worked on artificial neural networks but shifted focus to biocomputing, believing that the solution for AI advancements lies in nature [00:18:22].