The AGI: what is it?

If we think that the Artificial General Intelligence is the LLM that commands robots, we can ask one question. How many things can we order those robots to make? Or, how many things do we make in everyday life? If we forget things like military robots and some special-purpose androids. We can ask how many skills robots must have. That they can clean our houses and go to shop for us. 

And how many things do we do in our everyday life? If we forget things like lunches or some other things, can we make a robot that takes the grocery bag from the shop jumps in an electric car, and drives home automatically? The Large Language Model, LLM, and robot can operate as chefs if we want. 

If a robot has a module that allows it to make food, and check things like how ripe the meat is, it can make automatically food that the client wants. The client must only order some food. Then the system searches the merchandise list for the food. And after that, the robot will take those things from the shop. 

The AGI means general intelligence. That can use everything. But then we must realize that there is no single person on Earth, who knows everything. 

For making things in physical words the AGI requires a robot. The robot is the tool that connects physical items with computers. But if we want to use AI to make things like nanotechnology, that means that nanotechnology requires the tools that are suitable for nanomachine creation. Those tools are different than some screwdrivers, that operators can use to fix the car engines. 

The AI and especially Large Language Models, LLMs are new tools. The route to the Artificial General Intelligence AGI might not be as long, as we think. The modular data structure where the LLM works with multiple modules is a route to the AGI is route to the AGI. In the modular data structure, every skill that the AGI has is in the pre-programmed module. This thing makes it possible to create the entirety that the user can operate through the LLM. 

The system might have modules that it uses to control cleaner robots. The cleaner and housekeeping robots can be the same physical tool. However, the mission modules determine the role of the robot. The system has modules that control the temperature in the house. The module controls robot cleaners. The other module is connected to the robot car, which transports the robot to the shop to buy everyday goods. 

The partial mesh network is an open structure. That allows developers to connect an unlimited number of full-mesh networks into that data structure. 

The thing in the AGI is this. The AGI learns through modular data and database structure. When the AGI gets a new skill it simply creates a new module. The system can use so-called double mesh networking. The partial mesh protocol can involve the internal structure of the full mesh networks. So each ball in the partial mesh networking diagram involves the full mesh network. The open structure means that the system and its abilities can expand as much developers or the structure itself needs. 

Determining the AGI is difficult because the AI must handle every problem that it faces. But the AGI is still only a computer program. The full mesh protocol there the center is LLM. The LLM can call the module to make an operation if it requires it. But that requires the existence of a module. 

Sometimes people compare the AGI with things like mobile telephones. How many applications do we use every day? Is it E-mail, net browser, and some payment solutions? So, even if we have an AI that operates with multiple tools, can we call that thing AGI?

And do we have skills and knowledge to benefit the AGI even if we have it? We can have a robot car that drives a human-looking robot to shop to get the pre-ordered things. The modules limit the skills that those robots have. That means the robot can make many things.