AI forces us to rethink what “thinking " means.

Sometimes people say. That we cannot ever make an artificial general intelligence, AGI. The argument is this. No human can do everything. So, because there is no general human. There is no AGI. But the problem is this: the AI learns from modules. We can take the tiny artificial intelligence, or limited artificial intelligence, and connect new entirety from them. But. Can be lots of limited AIs. Be the AGI? The tiny AI means. A language model is created for compact devices. Like automated translators. Those systems. They can act as loops. For larger and more versatile AIs.  More about that later in this text. 

When AI handles information with certain rules like mathematical problems, it’s. An ultimate system. When AI handles fuzzy things. If the rules are not so strict. Those systems are in trouble. The AI doesn’t have opinions. They cannot simply tell which color seems better, red or blue.  The AI gives very interesting answers. But as  we might know. It doesn’t have an opinion. It forms answers from statistics. 

Mathematical formulas. And many other things. But it doesn’t have opinions like humans. AI can mimic feelings, but it doesn’t have them. The AI can control robots that  “defend themselves”. Or AI can say “ouch” if we step on those robots’ feet. But. Those things activate only because we pressed some sensors. We could make some of those reactions by using switches that activate tape recorders. 

AI prompts us to rethink intelligence. We might think that knowledge is intelligence. We might think that the ability to connect information from multiple sources is intelligence. And then we face the thing. That even if we can connect all the information in the world, we might not be intelligent. This means that we can process information that we can see. Machines make that thing faster than we. But then, when machines must jump out of the box. And try to make something abstract. 

The AI has imagination. It can use simulations. In simulations. The user. It can use the spacecraft’s digital twins as an example. For docking and other maneuver training. Then the AI knows when and how to use the control rockets. In the automatic model, two AIs can operate in the virtual universe. That virtual universe can be in the third server's memory. This means that two AIs form a loop. And the third AI controls that process. 

The machine will make things better. Or they simply fail. The AI can give wrong answers for many reasons. The first is this. The orders. The user's input can be too fuzzy. The AI can also use the wrong dataset. And this means that the user must be careful about orders. 

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

It’s possible that feelings have one purpose. To make our reactions unpredictable. If we did not have feelings. We would make our reactions predictable. If we were to attack like wolves. We would do everything in the same way. Without feelings, we would always attack or escape in similar situations. But feelings make it harder to predict how we react in certain cases. There. Some hunter who is just a puppy. 

Or when hunters threaten our family. The AI always follows logical ways. Of thinking. The AI will never suspect that anybody will tell lies. The trained AI can search the net. Are there things like funerals? For the name of the person who tells about the death of a family member? Or are there some other public ceremonies for that person? But the AI can always say. “Oh my goodness”. And that means the AI can mimic feelings. But it doesn’t have them. 

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

The problem with that paradox is that the AI requires well-articulated orders. To make things the user wants. Well-articulated orders are not possible without knowledge of the topics. For using AI. The user must have some kind of knowledge of the topics. Without that knowledge. It's possible that the AI makes mistakes. The mistakes can happen because of the databases that the system uses. Are corrupted. 

Otherwise, we can ask one question. If we give orders to some. Other.A person to make things like painting?  Does that painter ever make a painting that matches our imagination? In the same way, are there mistakes that AI makes sometimes between our ears? Do we just search for mistakes because we want to be better than our creation?

The AI helps us to understand how we think. The human brain involves millions of neurons. Almost all neurons have a pair. And those neurons form the loop. The information travels in the loop. The purpose of pairless neurons is to stop that loop if those neurons are stuck in that process without the ability to stop. The third neuron orders them to stop the process. And give space to another process. 

There is a possibility. That tiny AIs can form those loops in the larger language models. 

Theoretically, those processes. In the level of two or three neurons. They are easy to mimic. But the problem is that. Our brains have so many loops. That they are hard for programmers. We can say that thinking means connecting memories to information. That. The senses transmit to the brain. The fact is that. Something is missing in our puzzle. The thing that is missing makes us intelligent. Computers can connect and process information. But sometimes those things make mistakes. That even a child will not do. 

When we train AI, we require a lot of data. The AI can train another AI. But normally, operators train the machine. But as we know. Normally, it changes all the time. This means that two AI’s can form the loop that allows them to train each other. But in that case. The AI needs information. And information that AI uses can be wrong. 

The AI can teach or train other AI. The system can use the algorithms for the same purposes as human programmers. The algorithms. They are not hard to program. The problem is that the AI requires so many algorithms. And the infrastructure that it needs. Requires many microchips. Of course, engineers can make microchips. Or microchip groups. Those that operate in the three processor groups. The requirement for three processors. Or. Pairless processor core architecture is this. The pairless core releases the endless loops from the  processor pairs. 

But that is one depth of thinking. When we ask something about the AI, we must ask something. From a person who looks at us in the mirror. We must ask: what can we expect from the AI? Should we always expect feelings when we ask something? If we ask something. From the AI. We cannot always expect. That AI connects feelings into its answers. There is not. Feelings that we can naturally connect with all things. That we can ask. If we ask AI to solve a mathematical problem. There are no feelings that we can connect with that thing. 

And AI has no feelings. It can connect some phrases. Like “how horrible” to things that we say. Or tell it. The AI can mimic feelings. But it doesn’t have them. We don’t know what role feelings play in our thinking. There is a theory that feelings are created. To give a surprise element to our reactions. If we have no feelings, that means we would act the same way in all situations. Feelings make our reactions harder to predict. If we were to attack like wolves. We would do everything in the same way. Our prey or other people will predict our reactions. But feelings make those reactions harder to predict. 

https://bigthink.com/philosophy/how-ai-is-quietly-changing-what-we-think-the-human-mind-is/

https://scitechdaily.com/even-gpt-5-failed-this-human-attention-test/

https://www.unite.ai/the-small-model-uprising-why-tiny-ai-is-outperforming-giant-language-models/

https://en.wikipedia.org/wiki/Artificial_general_intelligence