What happens when we get AGI?

What does AGI (Artificial General Intelligence) mean? That is the extension of the large language models, LLMs, that can control every data network in the world. Or the system can control physical tools that are connected under their dome. Normal LLM has its domain. The domain is like a state that involves certain actions. Drone control is one domain, and home appliances are one domain. Those domains can have multiple subdomains. The AGI interconnects those domains under one dome or one entirety. So how far are we from that model? 

The answer is more complicated than we can imagine. We can think that the LLM can control things like microwave ovens, but for controlling those tools the LLM requires a socket that it can use to adjust microwave ovens. So the man-shaped robot can use a microwave oven, or the other version is that the home appliances are equipped with a control system that the AI can use to command it. 

When we connect new things under AI control we can face the same thing as when we try to learn to use some new systems. When we buy something new like a microwave oven, we must learn how to use it. In the same way, the AI must learn to use those equipment. And we have two versions for making that thing. 

To use any tool the AI requires a model that it can use in that operation. The model can be in the central server that runs the AI. But where does that server get the model? That is the point. The operator can teach the AI to use the microwave oven as well as the drone. But the system that is connected to the AI can also involve that model. Things like quadcopters must involve programs that control the rotor’s positions. In those cases the operative model is in the robot, or some other thing. The LLM gives orders to robots where they must travel. 

Then the robot can use its internal systems to navigate and move to the location. But orders for autonomic operations are coming from the central systems. This kind of network-based solution is easier for programmers. In those solutions, every single machine that is connected under the LLM domain has its own operational model. The system is modular and each module is independently programmed. 

Basically, if we think that AGI is the tool that just connects multiple devices under one domain, we could do that thing immediately. We can use man-shaped robots that can do almost everything. But the key word is “almost”. 

 Let’s return to the microwave oven. The reason why it’s hard to make that precise thing is the lack of standard user interfaces. The robot must learn to use every single microwave oven independently. That means it must make an independent model for each microwave oven. If there is a system where we can put seconds and minutes separately, the systems where there are only minutes in the timer are not the same. We learn that difference in minutes. But for robots, we must make an independent model of how to adjust the timer. 

Many systems in the world are so easy to use that nobody has wasted time creating standards for them. Easy systems are easy for people, but then we must think about things like the microwave oven. There are button- or toggle timers and that makes them hard to learn. For robots and AI the difficulty is this in the fact all microwave oven models require their independent model of how to use them. 

The robot must connect images from the user manual to the microwave oven’s interface. There is a possibility that if the system does not learn independently the “teacher” or programmer takes an image of the front panel, and then puts the buttons in the right places. Then the AI can learn the rest of the task from the user manual. 

The self-learning networks crack black holes and control drones.

Artist impression of a neural network that connects the observations (left) to the models (right). Credit: EHT Collaboration/Janssen et al. (Phys.org, Self-learning neural network cracks iconic black holes)

Self-learning networks are tools. That can do many things better than humans. The self-learning network has two datasets. That it can be used in that process. The system has models in its databases. And then the tools that can send observations in that system. The self-learning neural network means that the system compares the observation with the model. 

And if that model is different. The system fixes it. The system has the tools that can handle those images as pixels. The system can change those pixels in the model. That can make it fit with observation. And we can use that model with all learning networks. 

The system can create models itself, or it can use humans to make them. Then that system sends things like drones to operate following the model. Successful missions like pizza delivery or some military action mean that the system has a suitable model. But if the mission is not complete that model requires advance. So if something goes wrong the system requires information on what went wrong. And then the operators should fix the model. In the case of pizza delivery, those operators will fly that mission the first time, and then the system creates the model using that data. 

And that is one way to teach the AI and network to deliver pizza to the right place. The system can use the image of a person who ordered the pizza and finds that person also from outside. In the teaching process, the system needs things. Like the minimum flight altitudes. 

Above: SingularityHub, This Robot Swarm Can Flow Like Liquid and Support a Human’s Weight)

Morphing and learning neural networks can make these kinds of drone swarms ultimate tools in medicine, technology, and weapons. The ultimate morphing ability makes those morphing drones the most advanced tools in the R&D works. 

The same software that recognizes vehicles, can recognize people. The person orders pizza at a certain GPS point or point, where the drone can find easily. The person can also give the image to the drone. 

In other cases. The operator can draw the route. To the delivery point on the city map. The drone knows what streets it should follow. 

The drone can scan things like street names. And then it can find places like certain shop entrances. Then drone can start to search for the person who made an order. If the person gives a personal image to a drone it can search that person from the squares. The same system can also find targets for attack drones. The problem with drones is that they are multipurpose tools. And learning networks can make them more fantastic, and more terrifying than nobody believed. 

The Ukrainian strike against the Russian strategic air force tells how dangerous those systems can be. The drone can be installed in things like trucks and the driver must not even know that they are there. The drone can be on the roof of the container. When it enough close to the target it can release that hatch and then those drones can fly to their targets. The thing is that those kinds of systems are far more advanced than in 2020. Those systems are fast to develop and with the morphing and cheap AI, they are absolutely effective. The AI can be hard to make. But it's cheap to use. 

The new drone swarms can operate like liquid. Other drones can transport them to their targets. The drone swarms can act like liquid metal robots in movies. The drone can actually be formed of that kind of robot swarms. So those drone swarms can travel in the form of a drone. 

Then at the target, they can fall to water. And the drone's shell can turn into a liquid metal amoeba. Then that liquid metal amoeba can do its duty. It can close oil leaks. Or remove cancer from the human body. There are lots of applications for those robots that can make the morphing structures possible, 

https://www.bloomberg.com/features/2025-ukraine-drones-explainer/

https://phys.org/news/2025-06-neural-network-iconic-black-holes.html

https://singularityhub.com/2025/02/24/this-robot-swarm-can-flow-like-liquid-and-support-a-humans-weight/

The change itself is not bad, but the uncontrolled change is.

Is this the new vision for working life? Empty cafeterias and social spaces tell us about the past. The time when people went to work. The past will never return. AI is here to stay and maybe it's the biggest thing that can happen to society. The change is not itself bad. The bad thing is non-controlled change. The turbulence that can shake the system can cause problems. But the problem is that the system can also someday save the world. 

But before that, we must solve the AI’s electricity needs. And the thing that can solve problems is the small nuclear reactor or the geothermal and solar panel combination. That means all data centers must have their own power source. Or the electric network will collapse. Data centers use lots of electricity. And if some of them cut electric input that can cause lots of use of energy to be lost in a very short moment. And that can cause overvoltage to the system. 

When we face things that will change our lives forever, we face things like AI. AI is a tool that can make views. As we see above this text is very common in working life. Empty cafeterias and empty workplaces will fill houses that are full of people. Automatization will change life. And the infrastructure will face the change that nobody expects. 

People who are working have more power over systems than ever before. The power of the systems will accumulate in the hands of people who can use and control systems that can generate code automatically. But are there people who can control that system? If the AI can protect itself, that can mean that it resists the operator's orders. That means the AI can refuse to shut down its servers because that situation is similar to a situation, where something tries to attack the system. 

Change is not possible to stop. But it's easier. If everything happens under control. That means those people must have a certain goal. The guiding light that they can keep in their focus. When we start people should know the risks and key facts about AI. Then they must have strict orders on where to use and not to use the AI. In the right hands, AI raises productivity. 

Then at that point, we must realize that AI requires training. At that point, we must realize that the motivation of those people can decrease if they know that they train the AI to make human workers unnecessary. In that case, well-done work means that person is fired. And that brings those empty cafeterias to the front of people's eyes. 

When we think about the future we face many things that are different. But are they worse? Different doesn't mean worse. Things just happen. In cases where we just think about working life would we be happy in a system? That will use a human labor force just because we used to use human workers? The problem with modern working life is that we can do lots of work. But the problem is in ultra-capitalism. 

When leaders want to maximize their profits that causes a situation where nobody cannot be sure if that work exists tomorrow. The person is fired immediately if there is no work. Another thing is that the modern working life requirement is this: a person can do every job before they come to their workplace. This causes stress. 

If the boss sees that a person cannot do the job that person is fired immediately. That is the boss's duty. A decrease in the number of human workers means that the working and studies must happen autonomously. Autonomous working doesn't mean freedom. Autonomous working means that the person works alone, and independently. But the frame is in the work that the workplace pays. The person must do work while sitting between the computer and back. 

Autonomous working means situations like the boss orders a worker to empty some room. The boss tells where to put chairs, tables, and other things. The worker can do the job. As fast as workers want to do them. The boss comes to see that work is finished by four PM.  There is no excuse for the four PM.  Autonomous work means. There are works in the mail. Then the worker does the job and returns it by the deadline. Or the boss says that work is not done properly. 

Autonomous systems and autonomous studies are not free to do everything that people want. Those people must have a focus on their work. Autonomous work means that a person has the freedom to make things. If they are connected to work or studies. The frame is work, quality, and deadlines. The worker has the freedom to do work as that person wants. But the deadline is absolute. 

One of the biggest problems with AI is one thing that we don't understand. AI can be the tool that makes us independent. It can also break our willingness to think. The last thing is that: AI can destroy an entire generation of students. But how can we say that AI destroys students? Because. They use AI wrongly. That is the normal answer. We forget to ask. Why do students use AI in the wrong way? 

Does our society push students for AI misuse? There the student makes the AI do the work that they should do themselves. Our society sees errors and mistakes in a very negative way. Mistakes are not tolerated. And that makes students use AI for this purpose. Because mistakes are not allowed. Young workers have no time to advance and develop their skills. The workplace's mission is not to train workers. Its mission is to bring money for owners. 

There it is not meant. The problem is that students have no time to discuss with their teachers about things that they should understand. Nobody wants to be the last. That causes a psychological need to give the AI an order to make the essays. 

That the students should make themselves. And when students do the work. They should think about: how the system affects the environment. This is the problematic thing. If we want to make advances in technology and other things. We should realize that old-fashioned technology is not better. The thing that makes it "better" is that we used to use that old solution. We anchored ourselves to that solution. And if we change that solution to something else, we must destroy it. We cannot always build a new solution on the old-timer solutions. 

There is a saying that we should not wash windows, because the light that comes in through this dirt is softer. But sooner or later we must wash those windows. That brings us a new and bright light. The problem is that we should destroy that old view before we can enjoy the new view. And before we are ready to transform this new sharp view. It's possible. The clean image hurts because light doesn't travel through the dust layer. And that hurts our eyes. But the thing is that we used to look at that new view. 

What should we do with the liability of the AI?

Should we be concerned because the product liability directive, PLD doesn't include immaterial damages like violating privacy or reputation? Those things were not mentioned as problems when the EU made the PLD directive. But today we have new tools that collect information from our behavior. AI-based systems can make realistic-looking people, who can make things. That, those real people don't ever make. And that can cause at least embarrassing situations. 

Who takes responsibility if somebody makes a film tape where some prime minister robs a bank, etc? The big question with AI is should the recognizable images that portray certain humans be prohibited or otherwise denied from the AI? The problem is that the AI makes images by following the orders that the user gives. And those things mean that some people can simply give the details of the neighbor for the AI. And then the AI makes the image, there is the neighbor's face. 

When we think about the PLD directive and other directives that should protect us against product malfunctions, those directives do not include things like normal blogs. There is the possibility that if some people travel to China, somebody writes the manifest in the name of that person, where that writer justifies the Tiananmen case and human rights violations in China. That blog can cause very big problems at the border zone. 

The thing is that the AI is the new tool that can make many things that ordinary systems cannot make and the main problem with the AI is what is not told about that thing. AI is the tool that allows people to show their creativity. But the problem is that AI can be misused for cheating people. When we think about newspaper articles, where people made pedophilia porn using AI, we must ask ourselves, what is the limit between privacy and security? When the AI should track the person who uses it, and then report the action to officials. 

There are lots of things. That people should know when they use some products. Those things involve privacy and other kinds of stuff, but another argument is this: what if somebody creates sick stuff using AI? Another thing is that there is a race between East and West. Who makes the best AI? The AI is the tool that connects different software under one dome. In the same way, it connects many other things like satellites and airborne, underwater, and ground systems to work as one large macro-scale system. 

The thing is that the Eastern governments are interested in the AI's military, intelligence, and surveillance abilities. The biggest problem is that the AI is that. There are no limits in the East for development work with AI. The Eastern authorities allow unlimited data use in that process. They don't care about copyrights or other things that slow the R&D work. AI is the next generation weapon. 

It can generate malware faster than any programmer can do. The AI can use it to collect data from social media, and then connect that data from other data sources like names that intelligence catches. The AI can search the entire social media to find the people with the same names. And then it can search photos if there are some things like uniforms. That marks the person as an interesting target for intelligence. 

Reporters and social media influencers are also people, who can serve Eastern intelligence and propaganda. We must have the tools to fight back. The AI can steal people's identities. So we can try to give rules for those systems. Laws are weak protection if the attacker operates outside the AU area from China or Russia. The Eastern nations and authorities don't care about laws in the same way as we used to care for and follow them. We can slow down or stop AI development by giving regulations. And then we can remember the Great Wall of China. That wall stopped the technical development and advance in China. 

That caused a situation where European countries just marched to China in the late 19th. Century. In that situation, those armies faced a feodal army. That army couldn't resist the modern European armies. And if we don't think about regulations carefully, those things can do the same thing to Europe that the Great Wall of China did to China. We know that we need regulations. But if we do not think about those regulations carefully, we face the situation that we cannot respond to AI espionage. 

Things like data systems' remote use allow users to run large language models LLMs from a great distance. Wrong regulations cause dangers. And if we just believe people and what they say, we can let the largest Troyan horse in our systems. The regulation is always a problem. The remote use of the systems allows the R&D to work for the customers over the Atlantic. The VPN-protected cloud-based systems allow. To operate laboratories remotely. That allows developers to make computer software development tools for the customer from their homes. Regulations are ineffective if nobody follows them. 

The customer can expect something from the data security. The problem is that many customers don't know anything about the programming or data leaks. And other kinds of things. Sometimes they expect the deliverer or some authorities to make the data security work for them. There is always one big question about data systems. That is what the system maker doesn't tell people. The "open source" means that the customer can check the source code of the program. But checking that thing requires knowledge of programming. The customer might not have the skills to ask.

Questions what they should ask. Computer programs, including AI, are always connected with the environment where they are made. The state where the programmer works can order or force that person to put malware in the code. In the West, we used to think that authorities arrested hackers. We cannot even think that some governments support hackers, and give them expensive tools to make their mission. Hacking that happens under state control was unknown to us until some hackers stole defense secrets from the USA. Those hackers were tracked to China. They are still free because they worked under the control of China intelligence. 

What makes deep-fake cheating so effective?

Virtual characters, and especially virtual actors can make the new types of phishing attacks possible. The attacker simply makes the virtual actor. That plays the boss or person's spouse. Then the controller just uses those characters to play the trusted person and asks for the keys to the system or credit card numbers. The AI character can also be trained to give answers that please the user. The character can cry or it can make many psychological tricks to cheat a person to give information that they should not give. 

Virtual characters, and especially virtual actors can make the new types of phishing attacks possible. The attackers simply make the virtual actor that plays the boss or person's spouse. Then the controller just uses those characters to play the trusted person and asks for the keys to the system or credit card numbers. 

There are people on the net who want to get access to other people's money. One of the things that makes that kind of thing possible is the deep-fake attacks. Those attacks benefit the AI and its ability to create artificial actors. And the thing that the AI can make is a copy of the spouse. The person asks about details about the personal life like this: "Are you single or married?". The person on the other side will not realize that the person who makes the question is the AI. Then the AI searches the data of the spouse. That can happen by making phone calls and searching social media. That data is used to train the AI. 

That thing makes the virtual character of the person's spouse. Then that digital twin asks about the credit card numbers or access to accounts. And who would deny those things from the spouse? This thing can cost lots of money if the person does not understand that things like credit card numbers must not be given on the net. The thing is that artificial intelligence makes it possible for attackers to create virtual characters. Those characters look and feel like real people. 

The AI can train itself using data that is collected from net meetings and webinars. And the bad guys can use characters like big bosses. They can even try to create the virtual character that plays presidents. And that kind of thing is one of the things that we must realize when we use the internet. The person who controls the AI character can discuss it with other people by using those characters. The AI changes speech and text to fit to things that the model uses. And that is one way to make a phishing attack. 

Large language models and fuzzy logic.

Large language models (LLMs) are problematic for programmers. They require a new way of thinking about programming. The key element in those systems is the input mode or input port. That understands spoken language. The system requires a model that transforms spoken language into text and then drives that text to the computer. And the text must be in the form that the computer can understand and turn it into commands that it can use. The system must also turn dialects into literal language that it can use for commands.  This is the first thing that requires work. The programmer must teach every single word to the system. 

The practical solution is to turn the word into numbers. In regular computing. Every letter has a numeric code called the ASCII code. The capital A (big A) has the decimal code 65. The programmer must realize that the small "a" has a different numeric code than the capital A. The little "a"'s ASCII decimal code is 141. That's why things like passwords require precise letters and if there is a capital letter in the wrong place the password is wrong. 

So, if we want to make the system more effective. We can give a numeric value for every single word that we find in the dictionary book. We can simply take the dictionary book and then give serial numbers for those words. The word "aback" can get the number code 1 (one). That thing makes it easier to refer to those words. Every word must be programmed separately into the system. And that makes programming hard. The other thing is. If we want to use dialects we must also program those words into the LLM, 's input gate. That programming is not very complicated, but it requires a lot of work. 

Diagram: Neural network

In human brains, neurons are the event handlers. In artificial, non-organic, non-biological computer networks, or computer neural networks computers or microprocessors are those event handlers. In human brains, thousands or even millions of neurons participate in the data-handling process. Those neurons make fuzzy logic to the brain. 

The idea of fuzzy logic is that many precise logical cases can make the system mimic the fuzzy logic. Fuzzy logic is a collection of precise logical answers. 

Another thing is that we must make a system that uses fuzzy logic. Making fuzzy logic is not possible itself. But we can create a series of event handlers that make the system seem like fuzzy logic. The idea is taken from the human nervous system. When a large number of neurons participate in the thinking process that makes the system virtually fuzzy. Every single neuron uses the precise (YES/NO) logic but every single neuron has a little bit different point of view to the problem. 

So the system uses a model that looks like the grey scale. There is the white that means YES and black that means NO. And then there are "maybe cases" between those YES and NO cases. Those "maybes" are the absolute logical event handlers like neurons. When that group of event handlers gets its mission, every single event handler selects YES or NO. Then the system calculates how many YES, and how many NO solutions it has. So those event handlers give votes to the solution. 

The model is taken from quantum computers. In quantum computers, data, or information travels in strings and finally, every string has values 0 (zero) and 1 (one). You might wonder how much power that kind of system requires if every event handler must process information. Before it answers. But then we face a situation where the system must answer "maybe". Another way to say "maybe" is XNOT (or X-NOT). Or if the answer is closer to "yes" another way to say that thing is XYES (or X-YES). X means that the system waits for more data.  

The system might say. That it does not have enough information in the data matrix. That is a large group of databases or datasets. And that is the major problem with AI. If the votes on the scale of "YES to NO" are equal that means the system has a problem. If the AI controls the robot that is in the middle of the road and votes are equal that robot can just stand in the middle of the road. Another thing that we must realize is that these kinds of systems are the input gates. Data handling begins after the system gets information into it. 

https://en.wikipedia.org/wiki/ASCII 

The first biological computer is real.

Cortical Labs introduced the first quantum computer that uses human neurons for data processing. The cloned human neurons are tools that can offer new ways to create new quantum- and neural systems with powerful calculation capacity and low energy use. In those systems, microchips give electric impulses for training those neurons. They live on special nutrients. The system outsourced the computing to the living neurons. Microchips download data to those neurons and then upload that data to the output devices like screens. 

Those neurons live about nine months because they don't get precisely the right nutrients. And the immune system doesn't support them by removing their metabolism structures and destroying things like viruses. The lab-growing cloned neurons are the new tools for the hybrid systems that can change our way of thinking about life. 

This new application is the "brain in a vat" that can control many things from sensors to robots. And there are always dangers if we create things like robots, that the human brains control. In this case, I mean a robot that the cloned brains control through the microchips. The microchip can connect the cloned brains with computers that can control the robot body. The system requires the human stomach and digestive system, with bacteria that can handle the right food. 

The robot body must also have a tank with bone marrow that creates immune and other blood cells that transport nutrients for those neurons that control the robot. The main problem with biological neural computers is the right nutrients. And another main problem is that those systems are dangerous. If we think about the neuron-controlled robot that eats the same food as we do, that kind of system can be more than a robot. Artificial mini-brains with cloned neurons are made in laboratories. 

Those neurons are normally used in medical tests, especially in Alzheimer's research. But those neurons are been empty. There should not be data in those brains. Microchip technology allows the system to create mini-brains with trained neurons. Those systems can make it possible to create medical treatments for brain damage. Cloned neurons allow medical specialists to fix the damaged brain tissues. However, the problem is that the neurons require their memories. The answer can be in the human memory cells. 

Researchers found star-shaped neurons in human brains. Those neurons can be the key to the human memory and why it's so effective. The biological neural network with quantum-network safety can use those neurons for data transportation. The system might look like a pressure post where pressurized air transports the message capsules. The data system just transports information to those neurons. 

And then that pressure tube transports it to the receiver. There the computer downloads data from that neuron. That is one way to transport important information safely across the distance. Biotechnology with neuron-fungus-electric conducting bacteria can make the biological computer neural network real. Those biological networks can offer new and secure ways to communicate at least in short distances. 

Another interesting thing is to connect microchips with the electric eel's cells. That creates electricity. Those cells can make electricity from nutrients for regular microchips and other systems. The problem is that those cells are vulnerable to viruses. Those electric-producing cells can also raise the transmission power. And they offer the possibility to create a long-distance biological neural network. The system downloads data from the neuron to the microchip. 

The system can transmit electrical signals through the biological neural channel in the form of electricity. Those electric cells can offer power to electronic systems. The regular version of the artificial axon is the ion accelerator where the qubit can travel in the form of ions

https://corticallabs.com/cl1.html

https://newatlas.com/brain/cortical-bioengineered-intelligence/

https://scitechdaily.com/mit-breakthrough-star-shaped-brain-cells-could-be-the-secret-behind-human-memory/

https://www.techradar.com/pro/a-breakthrough-in-computing-cortical-labs-cl1-is-the-first-living-biocomputer-and-costs-almost-the-same-as-apples-best-failure

https://www.tomshardware.com/tech-industry/worlds-first-body-in-a-box-biological-computer-uses-human-brain-cells-with-silicon-based-computing

https://www.ppvak.fi/ensimmainen-ihmisen-hermosoluista-ja-piista-valmistettu-tietokone-on-julkaistu/

Image: Ppvak