“Two images from the Quijote simulations used in this study. The panels show the same region of the Universe, but in different cosmological models. The top image corresponds. To the standard ΛCDM, adiabatic cold dark matter model, while the bottom image shows a universe with massive neutrinos and modified gravity. “(ScitechDaily, AI Learned the Rules of the Universe and That Became a Problem)
The differences are subtle, but they reveal how changes in the underlying physics can affect the formation and distribution of cosmic structures. Credit: Francisco Villaescusa-Navarro (ScitechDaily, AI Learned the Rules of the Universe and That Became a Problem)
The term ACDM can also mean : the associated critical data model. That is the critical tool, when the sensor. It transmits information to the AI.
AI can help cosmologists, but it can also become a problem.
The method researchers call transferable learning can help them develop new models in cosmology and many other things. The term transferable learning. Means when the system learns something. It can apply. That learned thing. To other similar cases. So, when AI sees similar curves in some other cases. It can use things that it has already learned. To that other problem. This means that. The researchers must not always. Begin the training process. From the beginning.
The AI can search for similarities for the new thing in its memory. And if there is a match. That thing means that the AI. It can use that model for reaction. This should make AI more effective. The problem is this. The AI selects its sources using statistics. And that can make it hard to bring new data for the AI. Old research. They are very often-used sources. If somewhere is the new data. Before, nobody used the new data as a source. Old data dominates search engines. The AI is an excellent tool. When it must collect and analyse data from the galaxy movements.
But in cases like supermassive neutrons, the AI is in trouble. The AI is the best in business. When it must analyze precise information. Things like galaxy clusters and their movements are precise information. But in cases like supermassive neutrinos. The AI is not very good. At things where it must create models for new physics. When AI must observe phenomena. It can interpret them as the same. Even if they are different. Or in the cases.
There are some observations. Objects’ temperatures change. The AI might not know that the object’s temperature can change virtually. Because if something travels between the telescope and the object. That means. that the brightness or temperature. That reaches the observer changes. The AI might not notice things like clouds. In the Earth's atmosphere. Or other surprises when it observes some targets like Cepheid variables. If the system doesn’t know about that thing. It can recognize the Cepheid variable as a new star. If it doesn’t know that the star is a Cepheid.
When AI tries to analyze a certain point. That thing is very hard to do. But when AI must analyze. A very large entirety. The AI becomes more effective. The AI sees things. Like movements of galaxy clusters. And it can make. An analysis of the changes in those movements. We can use fuzzy logic to analyze how the star clusters move in the galaxy. But then we face a problem. If we try to predict. The movement of the galaxy. In its supercluster. That is hard.
We must know the entire system to make. A complete analysis with high precision.
The problem is in perspective. The thing that seems large on Earth. Seems very small in the scale of the Sun. And the sun seems very small in the scale of the galaxy. When the scale of the system turns bigger. The forces in the system are also stronger. In big systems. The phenomenon scale is larger. But they affect more slowly. From our perspective. The forces that travel between galaxies take millions of years to reach other galaxies. The distance between the Andromeda galaxy and the Milky Way. It is 2.6 million ly. So light travels 2,6 million years from that galaxy to the Milky Way. And that means that any force traveling between those galaxies needs 2,6 million years for that trip.
When we try to create a model. Of how one small sand bite behaves in a river. We must know many things. Like changes in the forces that affect the sand bite. But if we want to predict how the sand bottom behaves in the river. We can make that calculation very easily. When we think about galaxies. Stars are like sand bites on the bottom.
One star’s behavior is hard to predict. But the entirety is quite easy to calculate. And then we can go to bigger systems. In galactic superclusters, the galaxy is like sandbite on the bottom of the river. The force that affects the entire galaxy. Must be much harder than the force that affects sandbite. But millions of galaxies. They send. A very much. Energy. Many sudden things can happen in the galactic superclusters. Those events might not. Seem.
Like a very sudden thing. But an eruption in the core of the galaxy can start in milliseconds. Shockwave travels across the galaxy at the speed of light. So, if the star is at a distance. Of two light-years from the eruption source. The shockwave of radiation. It travels to that star. So, if Sagittarius A erupts violently in the core of our galaxy, the Milky Way. The radiation travels to Earth 26.000 years. The distance between Earth and that supermassive black hole. It’s 26.000 ly. The material, or plasma shockwaves, travel far behind that radiation shockwave. And the distance between plasma and wave movement increases all the time.
But. If things like supermassive black holes are in the trajectory. That makes them collide. That thing is very hard to change. When we face things like galactic superclusters. Things that happen on that scale seem very slow. But forces that put galaxies. To turn their trajectories into travel. At the speed of light. The force. That affects things. Like, turn their trajectories. Must affect a certain time with a certain force.
If we want to create an AI that analyzes galactic clusters star by star. We cannot make that thing. In the galactic scale, it suddenly happens. Violent eruptions. Those eruptions can break the entire model. In the scale of superclusters, events like supernovas don’t have enough force to affect the macrosystem. But a supernova could destroy things like dwarf galaxies. But if the supernova explosion happens in dense star clusters. That shockwave. Can. Launch other supernova explosions.
https://scitechdaily.com/ai-learned-the-rules-of-the-universe-and-that-became-a-problem/
https://en.wikipedia.org/wiki/Lambda-CDM_model
https://en.wikipedia.org/wiki/Sagittarius_A*
















