AI mimics human memory to make wise decisions.

"A recent study introduces the M2I framework, inspired by human memory, to address limitations in current large AI models such as inefficiency, high energy use, and lack of reasoning. By mimicking brain-like memory mechanisms, the research aims to create machines capable of continual learning, adaptive reasoning, and dynamic information processing." (ScitechDaily, Can AI Think Like Us? New Research Mimics Human Memory for Smarter Machines)

Human memory is like a puzzle. Every single memory unit holds one small part of the information. That the brain stores in it. The thing that makes the memory so flexible is that. There are millions of memory units. That brain can reshape and connect in millions of ways. It is possible to make this kind of memory for computers. Using the memory addresses. In those systems, we can say that memory allocation units behave like pixels in the CCD camera. 

The memory pixels that the AI can reshape and reorder make those systems powerful and flexible. But handling that memory pixel network entirety of the system requires lots of computer power. And advanced operating systems. The memory is more than image. It controls the physical and emotional reactions. When a thing like a robot carries packages the memory stores information on how it picks the merchandise up. 

A robot's memory orders it to move its limbs and react to things that the robot faces. When a robot sees something it searches match from its memory. If there is a match with even and description that description activates the action. 

Memory addresses can act like memory units or memory cells in human brains. Then, those memory addresses can make multiple connections of themselves. They can make millions of forms of the memories that the computers have. This kind of memory is like a CCD chip in the digital camera. The difference between the computer memory and the CCD chip is that the CCD chip turns the image into electric impulses. 

But, this network-based memory handles data stored in the system. The system's accuracy and flexibility depend on the memory unit or memory address's size and number of those addresses. The large number of small memory addresses or memory allocation units can create a more accurate and flexible memory entirety than the small number of large allocation units. 

When the system reshapes the image or puts those memory storages into a new order the small number of large memory allocation units makes the new memory image unsharp. The data that a large memory allocation unit stores is in fixed form. The problem with those data structures is the same as when we use a small number of large pixels in the CCD camera. 

That makes images unclear. The large number of small pixels makes those images clear. The same way the memory pixels act in the computer memory. AI thinks similar way as humans. It collects a group of data pixels and then reshapes them into a new form. 

https://scitechdaily.com/can-ai-think-like-us-new-research-mimics-human-memory-for-smarter-machines/