It's been a while since I posted anything on here. I've been busy with work and other projects, but I'm back and excited to share some new content with you all.
I recently watched a video about biocomputers, wait, bio-computers? hmm, interesting.
In one word i'd say 'Wild', this is not artificial intelligence, this is biological intelligence.
POV: some sand scientists just said, instead of AI, let's use brain cells, yeah yikes.
Brain cells are natural occurring units of intelligence.
Wait, did they kill anyone to get these cells? nah, let me explain.
Hmmm, they are created in a lab using a technique called human-induced pluripotent stem cells (IPSCs).
It begins with a cell donation basically, like skin or blood cells, which is then reprogrammed back into a stem cell and finally transformed into neurons, but how?
(IPSCs) Induced pluripotent stem cells are adult somatic cells (e.g., skin or blood) genetically reprogrammed to an embryonic-like, pluripotent state.
They basically just introduce master genes via vectors/viruses to overwrite an adult cell's identity. These factors* remove specialized chemical tags, transforming the cell into a pluripotent stem cell capable of becoming any tissue type, okay that's wild.
For context, this was discovered by Shinya Yamanaka in 2006, they can simply differentiate into any cell type. iPSCs are essential for disease modeling, drug discovery, and regenerative medicine without ethical issues, well let's see how they turn these lab create neuron cells into a computer chip, let's dive in.
////
The state of current AI: (0:00) Why brain chips? (1:17) DishBrain: (2:31) CL1: (3:45) Harvesting brain cells: (6:53) CL1 chip design: (8:16) Manus (Sponsor): (9:33) How it plays DOOM: (11:21) How it learns: (14:24) Benefits of living neurons: (16:56) Is it ethical?: (19:28)
Well back to brain chips, specifically a chip named CL1 that uses living human brain cells to perform computational tasks like playing the video game Doom. The presenter explains that traditional silicon chips are reaching physical limits, prompting researchers at Cortical Labs to develop systems using biological neurons, which are far more energy-efficient and capable of rapid learning.
Key Highlights:
From DishBrain to CL1: (2:31) The evolution from the original DishBrain prototype, which took 18 months to learn Pong, to the more efficient CL1 chip, which uses 200,000 neurons—comparable to a fruit fly brain—to learn complex games faster (3:45-6:53). Keeping Neurons Alive: (4:10) To maintain the neurons for up to six months, the chip utilizes a microfluidic perfusion circuit (synthetic bloodstream) to deliver nutrients and a filtration system to remove waste (4:41-6:13). The Human-Computer Interface: (8:16) Neurons are grown on a High-Density Microelectrode Array (HDMEA), which allows the computer to stimulate the cells with electrical pulses representing the game environment and record their neural activity to translate into actions (8:42-10:00). Training the Neurons: (14:10) Learning is driven by the Free Energy Principle, where useful actions (e.g., shooting an enemy) result in stable, predictable signals, while mistakes (e.g., hitting a wall) trigger chaotic noise, incentivizing the neurons to rewire for better performance (15:16-16:56). Future Implications & Ethics: (16:56) Living brain chips offer massive improvements in energy efficiency (running on roughly 20 watts) and could revolutionize robotics by handling unpredictable real-world environments better than traditional AI. However, this raises ethical questions about the potential consciousness of larger, scaled-up neural networks (19:28-20:49).
https://www.youtube.com/watch?v=ZqRtR6Z2U6U