The AI participates in new microchip development. New large language models, LLM, require a new microchip architecture that can drive and handle bigger data masses. When the AI drives a large data mass, the AI compiles the input data with databases. There are descriptions of how the AI should react to some situations. If details or parameters match with some database, that database begins a reaction.
The problem is that things like machine view require that the system analyze lots of data in a short moment. If the system has a neural network in use, it can share data with lots of computers. Image 2 is the neural network model and every single point between lines can be the independently operating microchip. The problem with a 2D neural network is that the data begins to travel from machine view through the system. In this linear model, the system shares data with processors one after one.
That a kind of linear model. Where data travels in a system like a wave. That system is quite slow if it controls things like robots on the streets. The need to drive and control large non-sorted, uncertain data mass requires new system architecture. The system should emulate human brains so that it becomes really effective.
Image 2. Diagram of the neural network.
The term "machine eye" means the input tool. That drives data to the system. In a quantum neural network, the machine eye's place is over the system. And it can drive data to multiple actors at the same time.
Then the delivery system shares the same data at the same time to multiple microprocessors. That means the neural network can start to drive all its databases at the same time.
Image 3. Diagram of the multicore processor where each box is an independent microprocessor.
The neural network will not stuck, because if there is no match between the input and the description, the database and its control system will crush the input. Modern microchips can have internal multicore architecture. Image 3 is a diagram of that multicore system that looks like a monolithic core. But inside the monolith is multiple microchips.
That architecture means that the microchip is actually a neural network. The internal network in the microchip means that the microprocessor involves multiple microprocessors. Each processor can operate as an independent microprocessor. Or they can combine their force into one entirety. These kinds of processors can revolutionize nanotechnology and things like drone swarms.
Comments
Post a Comment