Researchers are getting closer to creating a complete human brain simulation chip

We are getting closer to Bicentennial Man (a two-hundred-year-old, a famous American science fiction film) thanks to the breakthrough of neuron chip technology.

Researchers at MIT have finally found the key to creating a computer that works like the human brain - or at least close to the brain as much as possible.

In a groundbreaking study in the field of quantum computers, a team at MIT designed a chip with artificial synapses.

Before, in the human brain, synapses act as bridges for neurons to communicate by firing electrical or chemical signals through a cell receptor. It is a process that allows our brains to work and manage millions of tasks with ease. It is also how we make conscious actions and decisions and are thought to be the way we form ideas and thoughts.

In MIT's chip, creating artificial synapses allows precise control of the current flowing along them. This process mimics the way synapses regulate neurons in the human brain and is the key to the potential of building a complete artificial brain.

Currently, computers cannot compete with the computing power of the human brain - even if we suspect that Artificial Intelligence is smarter than all of us. In machine learning, trying to simulate all of our brain processes is a huge task. If you ask the AI ​​to perform some calculations while watching a movie render it can do it, but if the number is millions of small, continuous calculations while it will slow down the data collection process.

This is because computer chips operate in binary state, sending signals in the form of electrical impulses. The human brain is similar to using electrical impulses, but instead of binary signals, it uses synapses to regulate signals and activate the neurons needed to perform certain tasks.

Building this system on a computer has been an extremely complicated process. There have been many attempts but the brain architecture is too complex to create efficient simulation chips. MIT researchers have used two layers of conductive electricity separated by an amorphous "conversion layer" that acts like brain synapses to simulate this architecture.

A single chip can replace all these servers

Lead researcher Jeehwan Kim said: "When using a voltage to recreate data with an artificial neuron, you have to delete it and can rewrite it exactly in the same way. But with amorphous solid blocks , when rewriting, ions go in different directions because there are still many flaws, ionic currents are changing and difficult to control, this is a big problem - an abnormality of artificial synapses. "

MIT's human brain simulation chip uses a similar model, but uses silicon germanium mesh (Things you need to know about "Germanium" - Germanium) with one-way channels so that ions can flow through structural tunnels. structure instead of moving freely.

"In the end, we wanted to create a tiny chip like nails that could replace a larger supercomputer," Kim said. "This research has opened a breakthrough to create true smart hardware."

See more:

1. Overview of vulnerabilities on Intel, AMD, ARM chips: Meltdown and Specter
2. How to measure chip temperature, hard drive, video card . computer, laptop
3. Samsung produces dedicated chips to dig virtual money