Reborn in 2010, the AI ​​in my mind is too powerful.

Neuronal biocomputing chips are not exactly a new topic. Never underestimate the creativity of scientists. As early as 1972, scientists used planar microelectrode arrays to record the electrical activity of cultured cells.

Countless scientists have since attempted to create a stable platform for long-term recording and culture of neurons.

In 1988, laboratories had already established connections between silicon-based microelectrodes and neuronal membranes. In 1991, the Max Planck Institute successfully placed leech neurons directly onto insulated-gate field-effect transistors, officially establishing an information loop between cells and semiconductors.

After another ten years of research, scientists began to obtain some more interesting results, such as planting 50,000 embryonic rat cortical neurons on a 59-channel microelectrode array and connecting them to a robot to perform a simple approach-avoidance task.

In 2006, Japanese scholar Shinya Yamanaka published a groundbreaking paper in Cell, demonstrating that induced pluripotent stem cells could be obtained by introducing Oct4, Sox2, Klf4, and c-Myc, commonly known as OSKM Yamanaka factors, into mouse skin fibroblasts, proving that mature somatic cells could be reversed back to a state similar to embryonic stem cells.

In 2007, a laboratory successfully created human induced pluripotent stem cells by simply collecting some human skin cells using this method.

In 2012, just six years after its successful publication, Shinya Yamanaka won the Nobel Prize for this research, which shows how important this research result is.

Human pluripotent stem cells are not the same as the omnipotent cells found in fertilized eggs, nor are they the so-called original cells of life.

It can be divided into almost all cells in the three germ layers (endodermal/mesodermal/exodermal) of the human body, including nerve cells, cardiac muscle cells, and liver cells, etc.

Although this sounds incredible, it is actually not that difficult. Hundreds of laboratories around the world have already replicated it. Just by drawing some blood and skin, they can successfully differentiate cortical neurons on their own.

However, in most experiments, human neurons still failed to compete with mouse neurons, mainly because mouse neurons are more cost-effective. From embryo harvesting to digestion and dissociation, and even after two or three weeks when a synaptic network spontaneously forms and begins to fire, the survival rate is very high, and the cost of materials is only a few dollars.

However, human embryos cannot be directly extracted; they need to be reprogrammed using mountain methods. Directed neural differentiation takes four to six weeks, and maturation takes another four to eight weeks. It takes more than three months before they can be used, and there are significant batch variations, making the price ridiculously high.

For the resources on Lanai Island, price is not an issue at all; the only problem is time constraints.

Fortunately, some peripheral tasks could be outsourced to outsiders, and as long as the funding was in place, the quantity was not a problem at all. This project team, which Chen Yiran called the Top Player Project, began its rapid expansion.

Everyone in the project team contributed skin flakes and blood samples to begin culturing neurons.

Chen Yiran is ready to start working hard to advance the project. He doesn't want to wait foolishly for three months; he wants to at least test some things first.

So he chose to spend money, pouring money into relevant research institutes around the world.

Although people are now using technology from the mountains, they still haven't bypassed the technology developed by the University of Wisconsin in 1998 to isolate and culture human embryonic stem cells from human embryos. As long as the cells meet the characteristics, no matter whose method you use to create them, you will cross the patent line. You can publish papers, but you cannot commercialize them.

Chen Yiran secured the Wisconsin Alumni Research Foundation's support and ultimately obtained inventory from over a dozen institutions by donating to build a laboratory. Four weeks later, the project finally made progress.

Subsequently, the project team successfully began culturing neurons, which is very interesting. If neurons are cultured in a flat petri dish, they will obediently connect with the neurons next to them.

But if they are allowed to grow in three-dimensional space, they will begin to organize themselves. They will unfold, fold and layer, forming a rough prototype of a real brain region, somewhat like an embryo. These small spheres of tissue cells are called organoids.

Chen Yiran and his team cultivated 800,000 neurons, all of them human, and flattened the organoid spheres they formed onto a planar network of chips. In just two days, they successfully taught it how to play the pong game.

Although the health of some of the elderly people in the project has deteriorated, everyone is still very excited because this project is of great significance.

It's not just because this clump of cells is actually working towards a goal.

More importantly, this proves part of Chen Yiran's intuitive judgment, proving that these basic cells can indeed be mastered.

How do you teach a neuron? Chen Yiran chose a rather unconventional yet reasonable method: utilizing the neuron's aversion to chaos.

This is something that seems to be written in human genes; if you give neurons an unpredictable signal, they will feel stressed, and they don't like it.

So the setup is very simple, even a bit cruel. For example, in the most basic pong game where players catch and bounce the ball left and right, when they miss catching the ball, they are hit by a jumble of random noise.

When they successfully return the ball, the signal becomes calm and rhythmic.

These cells have absolutely no idea what pong is; they are not playing to kill time or to win, but rather they are trying to stop the chaos.

This method wasn't chosen randomly; it came about through discussions among a few old guys and is based on the principle of free energy. Essentially, this principle means that most living things always strive to avoid accidents and maintain the objectivity of their world.

Because being startled usually means something is wrong, these cells naturally have an urge to calm things down—the most basic thing living cells do.

Another month later, Chen Yiran and several assistants handcrafted an experimental computer in the lab, which they named T100. This time, they went a step further, putting the classic game Doom into the computer, laying neurons on layers of electrodes, and connecting the signals to movement and shooting.

When he performs well, the signals remain calm; when he is injured or even dies, they are overwhelmed by a large number of chaotic signals.

A week later, T100 learned the basic operations of moving and shooting. In the end, a bunch of neurons, a bunch of fleshy balls that would normally be completely incapable of thinking, used an instinctive ability to perfectly complete the game.

Afterwards, the project team connected the T100 to an AI all-in-one machine through a translator to transmit information bidirectionally, allowing the AI ​​program to monitor the game and decide when to reward and when to punish.

Then, something interesting happened.