Generative AI can turn SciFi into reality.

"Leading AI scientists warn of the significant risks associated with the rapid development of AI technologies in a Policy Forum. They propose that major technology firms and public funders dedicate at least one-third of their budgets to risk assessment and mitigation. They also advocate for stringent global standards to prevent AI misuse and emphasize the importance of proactive governance to steer AI development towards beneficial outcomes and avoid potential disasters. Credit: SciTechDaily.com" (ScitechDaily, When Science Fiction Becomes Science Fact: The AI Dilemma)

Artificial general intelligence (AGI) can be the large language model, that can create smaller specific language models. And maybe the computer or programming industry works around large-scale language models. The clients use the AGI to generate those limited AIs. 

The idea is that the customers pay to owners of the AGI for things like server maintenance. The owners of the AGI need money for electric bills. This is one of the reasons why researchers are working on things like geothermal and solar panels for green energy production for server halls. If server halls use green energy, they don't pollute. 

And maybe the question,  what the employer hears at the magistrate when they register a company is "Have you downloaded your own AI yet"? 

And then they can download the limited "baby AI" to their servers. The limited AIs are modules that can be created to handle larger-scale and more complex data structures. 

The main problem is that people like Chinese intelligence can use those limited AIs as modules that they can use to create their own AGI. The intelligence just creates multiple companies and then downloads the limited AIs to those companies. Then those limited AIs can act as modules for complicated data structures. 

The machine as intelligent as the human waits for ten years. This is one prediction for the future. When somebody would tell me that someday we can make that thing, I would say that this kind of AI needs so many databases that it's too hard to complete. But today generative AI can make those 100-200 billion databases, if it has space, where it puts that structure. 

The intelligent machines are frightening. Somebody resists them, but they are coming. One of the reasons why somebody resists those systems is that they take jobs from humans. The problem is that. This thing like unemployment interests people only when the AI takes their jobs. When the U.S. car industry took jobs to Mexico, nobody was interested. That thing caused unemployment and social problems. But the police were enough to handle that problem. The media is interesting only when the AI takes jobs from reporters. 

The AI can make life easier. It can make working days easier. And it leaves time for innovation. But unfortunately, AI is the tool, that allows to fire half of workers. And that is the problem. The company leaders don't see AI as a chance to be more generative and more innovative. They see it as a chance to decrease the costs and crew. That is one problem with companies. 

The companies like Space X are very innovative. They know how to benefit from the publicity that rocket engineering brings to them. And that is a good thing for Elon Musk's business. The reason why Space X detonates rockets in front of people's eyes is that: these kinds of things are interesting. And interesting things bring publicity. And maybe some companies will come to buy a launch from Space X when they see that work. 

That thing causes one problem. The space industry, and especially the space launch industry accumulates into the hands of companies like Space X. The Pentagon made contracts with Space X and Starlink about their communication, location, and other systems. That thing gives very much power into the private hands. That means the private companies will get the military forces' communication in their hands. 

That means the world is decaying. There are innovative companies that make many new things. That innovation brings big profits for companies that own the patents. And then there are non-innovative companies that pay for those innovative companies about their products. That thing makes Space X and those kinds of innovative actors more and more powerful. 

Generative AI is a tool that can revolutionize engineering. In the production stage, the engineers must only give the requirement specification to the computer. Then the AI selects raw materials. And then it can start to create the product. The industrial robots can have 3D printing systems in their body. And that makes those systems more flexible than we ever imagined. 

The system can follow the data flow that comes from the rockets. And by analyzing that data the system knows if there is a problem in a certain engine. If a rocket uses many small engines the system can shut down one or two engines, if there are failures. In that process, the system just cuts the fuel and oxygenize from the engine. 

In the case of an emergency, the system can just separate the upper stage. And try to bring the payload safely to the ground. That can happen using a parachute or if the upper stage can land vertically that thing can bring the payload safely to the ground. 

https://scitechdaily.com/when-science-fiction-becomes-science-fact-the-ai-dilemma/

Pulsed plasma rockets are an interesting solution for Mars missions.

"Howe Industries is developing a Pulsed Plasma Rocket (PPR) capable of producing 100,000 N of thrust with a specific impulse of 5,000 seconds, promising to revolutionize space travel by enabling faster and safer manned missions to Mars and beyond. (Artist’s concept.) Credit: SciTechDaily.com" (ScitechDaily, Mars in a Flash: How Pulsed Plasma Rockets Are Revolutionizing Space Travel)

The new pulsed plasma rockets (PPR) with 100,000-newton thrust and a specific impulse of 5000 seconds are the tools that can transport humans to Mars. In a very short time. If we compare that time with chemical rockets. The pulsed plasma systems can use fusion, fission, or antimatter to create high-energy plasma that pushes the rocket forward. The light antimatter system uses the antimatter or positron injection into the water or hydrogen. 

The system that raises the propellant's temperature can also be radio waves, microwaves, or lasers. In a radio wave-based system. Plus and minus radiowaves impact the propellant. And form an electric arc. In the microwave-based system, the engine heats propellant using microwaves, and then a magnetic field pulls that heated plasma backward. It's laser plasma engines. The system uses lasers to heat and ionize propellants. 

Simplified image of the Pulsed Plasma Rocket (PPR) system. Credit: Brianna Clements, edited (ScitechDaily, Mars in a Flash: How Pulsed Plasma Rockets Are Revolutionizing Space Travel)

It doesn't matter how the heating systems or ionizers get their energy. And that means the pulsed plasma engines can operate using solar power. Nuclear rockets always need long wings to decrease the reactor's temperature. So the system may use solar power for at least part of the mission time. 

Small-size pulsed plasma engines that operate near Earth can get the power remotely from high-power radio transmitters or laser satellites. That kind of system can transport humans between Earth's orbiter and the Moon. That means the moon shuttle can also use pulsed plasma. And we can say that the moon shuttle must not have the same capacity as Marscraft. 

The idea of those (systems is simple. A rocket engine raises the material's temperature to a very high level, and then that expansion pushes the craft forward. One of the problems with pulsed plasma engines is how to control plasma. If high-energy plasma touches the plasma channel wall, it burns that will immediately. 

https://scitechdaily.com/mars-in-a-flash-how-pulsed-plasma-rockets-are-revolutionizing-space-travel/

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

The AI is an excellent spy.

The AI can lie if programmers order it to do that thing. One of the biggest threats in the AI world is that somebody creates a copy of the well-known AI. And then, that actor turns the data traffic to that trusted AI's digital twin. The limited AIs can used as the modules that operate under one domain. The versatile AIs can act as attack tools. 

Or the digital twin of the well-known AI chatbots can used as the greatest honey pots in the world. The AI-based chatbot can store all queries that users make. And answers that they get in the mass memories. There the intelligence can analyze that data. 

The attackers may work for some governments. And if they can create a large number of limited AIs they can create the entirety that is the same way versatile as the Chat GPT or Bing. But the programmers can modify that thing into the cyber attack tools. 

When we talk about language models and especially network-based giant applications. There is always a small possibility that hackers create duplicates of famous artificial intelligence tools. In that case, the hackers can create the language model, that generates new types of malicious software. The hackers can double the data that travels in and out from the language models, and then they get confidential data about things. That is not meant for the public. 

Another problem is that it's possible to create the digital twin of the language models. The operators can create limited AIs. And then connect those limited AIs to the new entirety. In that model, the smaller AIs are the modules in the system. And making smaller AIs is possible to create a system that looks like the Chat GPT or Bing, and then those people can route traffic to that fake system. Those systems can modified to attack tools that can break any firewall or antivirus. 

The AI doesn't think as we think. That means the AI can tell lies if programmers order it to do that. The AI's answers are programmed in its code. The programmers can order the AI answer certain way to the certain questions. 

If somebody asks the espionage AI does it make some kind of data fishing? The AI can say that it will not make that. Even if its main purpose is to collect data from secured systems. The AI can give false information if that action is programmed for it. The AI can also create fake memories for itself. When we think about drones and robots the AI can clean the mission recordings and then replace that data using some other drone's mission records. So the AI remembers what its users want. 

https://www.sciencealert.com/ai-has-already-become-a-master-of-lies-and-deception-scientists-warn

New AI-driven drones learn like animals.

"Photo of the “neuromorphic drone” flying over a flower pattern. It illustrates the visual inputs the drone receives from the neuromorphic camera in the corners. Red indicates pixels getting darker, green indicates pixels getting brighter. Credit: Guido de Croon" (ScitechDaily, The Future of Flight: Researchers Develop Neuromorphic Drones That Learn Like Animals)

The animals and their training are inspiring new drones and other robots. The idea is similar to teaching dogs. When a dog makes something as it should, the instructor gives candy to it. That thing motivates the dog, and in the case of a drone or robot, the operator pushes enter and accepts the operation. 

In the traditional learning model, the operator remote-controls the robot when it makes something for the first time. Then those operations are stored in the machine's memory. But the more advanced systems can imitate other machines or humans when they learn something new. 

In those systems, the robot observes other things like similar robots using cameras. And then, it copies those actions into its mass memory. If we want to make man-shaped robots, that imitate humans in the learning process, the robot can see how human moves hands and legs during some actions. Like moving boxes.

Then the robot can copy those actions into itself. Then the robot can use those movement series when it takes order to move similar boxes. The robot sees the box and then connects that image to the databases that involve the movement series. The robot can learn things by watching screens and in that model, the robot can sit in an armchair and look at the TV to learn things. Like how to make food. 

When a computer learns something, it connects a new skill module into itself. If we want to make a drone helicopter that must fly through the labyrinth we can use the film where a remote-controlled drone flies through the route. Then that film will driven to other drones' memories. Things like GPS, inertial navigation, and laser rangefinders can make drones operate independently. 

However, using drone pairs is possible to create similar results using cheap drones. That must not be equipped with complicated sensors. Only cameras are enough to make impressive solutions for independently operating drones. 

In the simplest and the most effective model drone pairs can operate as a team. The drones observe each other and tell if the other drone goes too close to the wall. The drones must only know the speed of the remote-controlled drone.

Then drones must copy the turnings to themselves. If drones are identical, they can use information on how many rotations per minute the remote-controlled drones' engines take. Then the drone needs time what certain part of the labyrinth takes. But if the labyrinth is open, another drone can hover above the second drone and tell where that low-flying drone must go. 

https://scitechdaily.com/the-future-of-flight-researchers-develop-neuromorphic-drones-that-learn-like-animals/

The future of work is cooperation with AI and humans.

The AI is the tool, that divides the opinions. Some people resist it because they are not ready for that. But is that the real reason for why some people search only negative things about AI? AI can deliver our resources for productive work if we want to do that thing. 

But it also can give the possibility to sack 50% of workers. Then we can think about this "we are not ready" argument. When we see people who say that "we are not ready for AI", we can ask "Who are we"? When workplaces took computers in offices. Nobody said that workers are not ready. 

Everybody was ready when computers came to their desks. There were courses, and then if the person didn't learn how to use a computer. Well, that person was allowed to find another challenge. And still today there are people, who are not be you with everyday tools like mobile telephones. But we must use those systems because they are so fantastic. 

The mobile telephone can break our privacy, and if we carry company telephones in our pockets, that thing allows the telephone's owner to see our location and what we talk about every second. So why do we carry our workplace's telephone in our pocket? If a company gives a telephone to us, that telephone should be only for work use. It is probably not meant for personal use. So why our workplaces don't lock those telephones outside the working day? 

Artificial intelligence is coming, whether we are ready or not. Those algorithms are tools that can make our lives easier, or they can turn our lives into hell. In the hands of authoritarian governments, AI is the ultimate and destructive tool, that can break privacy and make all mobile devices the surveillance tools. 

The deep fakes and digital actors created using dead bots can be used in intelligence work. But they are tools that can break people's privacy. It's possible that people make digital twins of themselves and then slip to the beach on a sunny day. And the digital actor represents them in the digital meetings. Those tools are fundamental things, and creative AI can make them operate exactly like humans. In the hands of hackers, those things are good tools for fishing rod information of people, who are not prepared for this kind of attack. 

The problem with AI is that it can also make many things more effective. Things like dead bots are tools that people can find on the net. And misuse of those things is possible. Things like FSB can use virtual characters to uncover things that governments cannot tolerate. But also Western law enforcement officers can use those tools to uncover pedophiles and drug dealers. So in this case the use of AI depends on the user.

And it can release our working time to make and innovate better work and products. When we see AI, we must see it in its entirety. Things like "Dead bots" and deep fakes are tools that can cause moral and ethical discussions. Why the AI don't mark the deep fake images and deep fake animations? Using marks they made using the AI? Deep fake animations are only a series of deep fake images, and those things are a threat to democracy. Deep fake animation can used to destroy people's reputations. 

Dead bots are dead people's digital models. But dead bots can used to make digital twins or digital actors using the data, that a person shares on the internet. That thing allows the people to make the digital actor that portrays themselves and that digital twin represents people in digital meetings on a beautiful summer day when a worker is at the beach. But those digital twins can also be used as espionage tools. Intelligence officers or hackers can use dead bots for phishing attacks. 

Attackers can create digital twins of the company leaders using dead bots. And then, they can start to keep the digital meeting. The AI and interactive actors make it possible. The digital actor can also make works that they get. And those things can make it exactly like real humans. 

The breakthrough makes it possible to create sustaining chemical compounds for nanotechnology.

"Researchers have detailed the structure and function of the enzyme styrene oxide isomerase, a tool that enables green chemistry by facilitating the biological equivalent of the Meinwald reaction. This enzyme’s ability to produce specific products with high efficiency and stereospecificity holds significant potential for the chemical and pharmaceutical industries, promising more sustainable and environmentally friendly processes. (Artist’s concept.) Credit: SciTechDaily.com." (ScitechDaily, Bionanomachine Breakthrough: A Master Key for Sustainable Chemistry)

Nanotechnology makes it possible to create new types of treatments. In medical use, long and complex molecules make sure. That medicine is released only at the right point. When an enzyme touches the medicine molecule it cuts chemical compounds. 

And that activates the medicine. The medicine molecules can equipped with the transportation molecule or enzyme that makes the medicine travel into the right cells. This molecule can be the nutrient that the targeted cells or bacteria eat. 

Nanomachines are perfect tools for things like medical treatment. The main problem in those things is how to control nanomachines. The nanomachines are complex molecules that act like machines. The complex molecules require mass production. 

Nanomachines can used in medical work. The problem with nanotechnology is how to make sustaining chemical compounds. The second problem is how to make nanomachines select the right cells. And the third problem is how to control those nanomachines.? 

And the second problem is how to destroy those molecules at the right moment. If we think of the wheel-looking molecular machines that are above the text, we must understand that the size of those machines is very small. The system can use acoustic whirls to press ions and anions together.

Nano-wheels can destroy bacteria in three ways. The nanorobot can slip into bacteria, and then it starts to rotate very fast. That thing destroys the bacteria's internal structures. The nanorobot can start to rotate very fast and make nanobubbles that can destroy bacteria. 

The fast-rotating nanomachine can form a so-called acoustic bubble. When that bubble starts to oscillate. It forms a low-pressure area around its shell. That causes liquid vaporization. And that forms new bubbles. Or the acoustic system can detonate the nanomachine, which makes holes in the cell's shell. 

The wheel-shaped nanomachines can used to create nano-size bubbles. The system just makes those wheels rotate so fast that they form supercavitation. Those fast-rotating wheels form nano-size bubbles in liquids. Those bubbles can used to block blood vessels. They can capture viruses or bacteria. And they can used to fill bacteria. 

Nano-size bubbles can used in medical solutions. Nanobubbles can close the blood vessels. Researchers can use them them destroy non-wanted cells. The wheel-shaped nanomachines can also create bubbles in the water, and they can destroy bacteria from dirty water, When those machines are not needed anymore, the acoustic system just makes them resonate, and that thing destroys the nanomachine. 

https://scitechdaily.com/bionanomachine-breakthrough-a-master-key-for-sustainable-chemistry/

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

A silicone offers a new way to make the quantum superposition and entanglement.

"Researchers at the University of Manchester and the University of Melbourne have developed an ultra-pure silicon crucial for creating scalable quantum computers, which could potentially address global challenges such as climate change and healthcare issues." (ScitechDaily, World’s Purest Silicon Paves the Way for Next-Gen Quantum Computers)

A silicone offers a new way to make the quantum superposition and entanglement.

Quantum entanglement does not necessarily happen between two particles. The quantum computer can make that process also between two electromagnetic or quantum fields. The vital thing in that process is that the quantum fields are slight enough. Then the system can make the superposition and entanglement between those quantum fields.

The photoelectric phenomenon makes it possible to transform the solar panels into the quantum computers. But a practical solution requires absolute pure silicone. In those cases, the laser ray transmits data into the silicone layer. The biggest problem with this kind of thing is that the silicone must be pure. And then the system must also stabilize the silicone.

"Scientists have developed a new technique using ultrafast terahertz pulses to control atomic motion in two-dimensional semiconductors, promising advancements in high-speed computing and electronic device development." (ScitechDaily, Scientists Discover New Semiconductor Excitation Technique)

For making silicone-based quantum computers, researchers must purify silicone chemically and physically. "Natural silicon is made up of three atoms of different mass (called isotopes) – silicon 28, 29, and 30. However the Si-29, making up around 5% of silicon, causes a ‘nuclear flip-flopping’ effect causing the qubit to lose information." ScitechDaily, the World’s Purest Silicon Paves the Way for Next-Gen Quantum Computers).

The system must create monoisotopic silicone. That thing can created using centrifugal separation, which is used in the nuclear enrichment process to separate fission-isotopes from non-fissile isotopes. Same way. As a centrifugal separator can separate U-235 from U-238 a similar system can separate the silicon 28, 29, and 30 from each other.

Pure silicone offers the possibility, that computers can manipulate the quantum fields using laser light. The system requires pure silicone because that thing removes the quantum noise from other atoms. And that makes the silicone atoms exchange information in that structure without disturbing signals that can disturb quantum entanglement.

The purest silicon in the world paves the way for the next-generation quantum computers. Silicone is a semiconductor that has a photo-electric phenomenon. This ability can make it possible for the system. That it can use the light to manipulate the qubits. In this kind of qubit system, the electric phenomenon forms the "energy hills" in the structure. And then, the system puts those energy hills in quantum-level superpositioned entanglement.

In some other visions, developers put two 2D silicone layers against each other. Then the system makes quantum superposition and entanglement between those layers. The silicone layers can be separated using graphene. The thing is that there are many ways to make the quantum superposition and entanglement. In some ideas, quantum computers use the atom's quantum fields to make this effect. In that case, the superposition is made between two small-size quantum fields.

https://scitechdaily.com/worlds-purest-silicon-paves-the-way-for-next-gen-quantum-computers/

https://scitechdaily.com/scientists-discover-new-semiconductor-excitation-technique/

https://learningmachines9.wordpress.com/2024/05/14/a-silicone-offers-a-new-way-to-make-the-quantum-superposition-and-entanglement/

Does the universe stretch or does it create space?

The universe's expansion is one of the strongest evidence of the Big Bang. The Big Bang was not the bang. It was an event where energy and material reach a certain energy level. All known materials and energy formed in that case. And the problem is that the universe's expansion is not what it should be. The problem is that dark matter and dark energy dominate the universe. In some models, dark matter could exist before material formed. 

Or material reached energy levels, between our universe's energy minimum and energy maximum. But if the dark matter existed before the Big Bang. And its temperature is lower than the energy minimum in the universe, which means visible matter covers that dark matter by its energy. Cold Dark Matter (CDM) may consist of a material whose energy level is lower than the energy minimum in the universe. 

In some models, there are two forms of dark matter. 

Does the hypothetical dark matter particle called weakly interacting massive particle (WIMP) be colder than absolute zero, or is its temperature something like <0,5 kelvin? Or is its spin so fast that it can make energy or wave movement travel past it? And could we someday produce dark matter? 

1) Cold dark matter

Cold dark matter that temperature is lower than zero kelvin degrees or just a little bit above zero kelvin degrees. The absolute zero point or zero kelvin is the temperature where all movement ends. It's the energy minimum in the universe. 

But outside the universe is lower energy levels. That matter is dark because the energy travels in it. And then the energy level reaches the energy minimum in the universe that matter existence ends. 

The reason for that is that the existence of a particle requires it to have a different energy level than its environment. If a particle's energy level with its environment turns the same, the particle merges into wave movement. 

So the hypothetical weakly interacting massive particles (WIMP) can be unseen if their energy level is lower than zero kelvin. Or actually, the WIMP energy level can be lower than 3 kelvin radiation. If a particle's temperature is lower than the cosmic background it pulls energy into it. If the reflection temperature or energy level is also lower than the cosmic background the matter is not visible, 

2) Hot dark matter.

Hot dark matter. In some models, the spin of those particles is so fast, that they throw electromagnetic fields past them. In that case, the high-speed spin denies the reflection. And wave movement travels or passes those particles like radiowaves pass the stealth aircraft. 

The fast spin creates a standing wave around those WIMPs because outcoming radiation cannot penetrate them. The radiation that reflects from WIMP impacts with outcoming energy forming standing waves.  Then that standing wave makes outcoming energy travel around that particle. 

In the next dark energy model, the source of the dark energy is in the electromagnetic whirl that closes the universe inside it. That energy field and whirl are invisible to us because their energy level is lower than the universe's energy level. 

If there is radiation outside the universe, we cannot see it. If radiation and particle's energy level is lower than in the universe, energy travels one-way. And because energy travels out from the universe, the outside world or dimension outside the universe remains invisible. Even if there is a reflection that comes out from that material. That reflection must have so high energy level so that it can penetrate the Universe. 

But then we can think of the universe as a bubble. In some models, the universe is a bubble in a giant electromagnetic tornado. The tornado would be in the wave movement the energy level is far lower than in the universe. That hypothetical tornado or tunnel is not a wormhole. It's not a tight structure, and it lets energy travel outside the universe. It's the electromagnetic whirl that sends energy through the universe. And then that energy reflects from the universe's center. That forms the dark energy. 

The Universe is a bubble that exists in the wave movement. The wave movement's energy level is lower than the universe's energy level. If it's outside the universe. The universe like all particles in it is oscillating. And that oscillation forms electromagnetic vacuums. And then there are form new bubbles and whirls in the energy field. 

If we want to model that hypothetical phenomenon in the chemical- or molecular world, we must know that. Acoustic bubbles can multiply themselves. When the acoustic system oscillates those bubbles, their membrane starts to oscillate. The fast-moving membrane forms low pressure in the liquid. And that thing can boil the liquid. And that thing causes a situation in which acoustic bubbles form new bubbles. 

https://bigthink.com/starts-with-a-bang/universe-expand-stretching-creating-space/

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

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

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

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

https://en.wikipedia.org/wiki/Lambda-CDM_model

Researchers can use AI-driven video games to test machine learning.

Statistics-based machine learning is quite easy to make. 

AI playing video games is a big deal because those systems can used to create machine learning. The virtual environment is an excellent place to develop AI and its ability to learn things. The screenshot above this text is from the sniper game on the Y8 homepage. In that type of game, the AI can observe the direction in which the gamer moves. 

The machine learning base is in statistics. The system makes a database about the mover's directions. Then the AI opponent will use those statistics to aim the gamer. Those statistics are the tool that will make the AI an ultimate opponent. 

But that is the biggest weakness of this kind of statistics-based algorithm. There is the base model about the direction where the gamer wants to move. Most gamers will turn in a certain direction. Because they are right-handed. Left-handed people are better, especially in martial arts like boxing. The reason for that is that the punch comes from the wrong side. If the opponent is right-handed. 

If the opponent gamer is left-handed the system will make the wrong estimation about the direction where the gamer moves the character. Most people are right-handed. And most right-handed people follow certain formulas to select the way, where they want to move their character. The left-handed person thinks oppositely to the right-handed people and that means the AI aims in to wrong direction.

The next-generation autopilot parking systems. And use of virtual reality (VR) to test AI-based autopilots. 

The drone connected to the next-generation autopilot allows the AI can drive the car into the parking very effectively. The drone can be in the box on the roof of the car. When autopilot drives the car to the parking, the drone will rise above the vehicle, and then the autopilot can use that drone's camera to see, where the car's corners are going. 

Computer games can used to test how fast people can learn. The game is a virtual environment. And it can used for many purposes. The virtual space allows the testing of self-driving car's visual observation systems. The car can be on rolls, and the system uses interactive movies to introduce different situations for the car. The system can use media projectors to make the VR environment. 

The system follows the car's and its autopilot's reactions when its camera system sees something, like a moose that suddenly comes to the front of the car. In those cases, the vehicle's camera system drives information into the AI-based autopilot. During the R&D process, the developers can control the vehicle's digital twin in the real world. in that process system collects data for autopilot's program. 

Then collected data will first be stored in the autopilot system that is in the VR room. There is a safe environment developers and AI adjust the system so that it will be as safe as possible. 

https://bigthink.com/the-future/why-ai-playing-video-games-is-a-big-deal/