The AI can have a physical body. The robot body communicates with supercomputers. And it makes them more flexibility.
AI learns the same way as humans. The learning process and its power depend on the diversity of the information. AI requires versatile information from multiple sources. And when we think about AI. And its ability to learn things.
We must think about it. Why and where we learn. We can try everything ourselves. But there is another way. We can network with other people who do similar things. And then we can share our experiences and thoughts with other people in that network.
We might have a good education. But we go to learning meetings. We learn from other people's experiences. Sharing information makes networks effective tools to learn things. In that model, the single actor must not make and know everything.
When we talk with other people we can expand our view of things. We get more ideas when we meet other people and share our thoughts.
We can work with those ideas. And mix them with our environment. That thing extends our corridors and predisposes us to new information. Information plays a vital role in the learning process. If some actor in the network, which can be human or some server faces something. That thing can be shared with another network. If the server is under network attack the system can collect all data from that event into its memories. Then it can share that data all over the network. And other actors can mimic that server to defend themselves against similar attacks.
Similar way. AI should talk about other things. If the AI is just a language model. It has limited ways to learn things. Mainly large language model learns in a verbal way. And that is a very limited way. It's easy to write things to LLM and order it to do something when something happens. This can be enough in cases where the AI should detect and defend against network attacks.
That program creates a reflex to the system. When something happens that system reacts like it is programmed. Think about a case in which you should explain everything using words. That thing is possible. But it's more limited than if the AI can use images or films in the learning process.
That means the AI can learn things from the homepages. And maybe from surveillance cameras. But if the AI has a physical form like a robot that interacts with a server, that runs the AI that extends its ability to learn things. The AI learns things visually. By connecting certain images or things with certain actions. That is a more versatile and easier way to teach things to AI. The physical body that communicates with the server can be discussed with people.
The robot body can keep in contact with the LLM. The system can operate remotely. The LLM works in servers or in morphing neural networks. Those servers can be in the bunker. Or the system can use non-centralized computing. That means the system can share responsibilities all around the robot groups. The system just connects robots computers into entireties.
In some futuristic visions, humans will fly to the Kuiper Belt to make quantum computers in that cold and stable environment. In the Kuiper Belt, every metal is in superconducting condition. So that means even human-size robots can have quantum brains. That gives them extremely powerful computing capacity. The low temperature with a static environment makes the Kuiper Belt a promising place to make quantum computers.
https://bigthink.com/the-future/ai-cognition-and-the-road-to-meaning/
