Skip to main content

The model collapse can danger LLM.


An example of model collapse in an image-generating AI. (FreeThink,“Model collapse” threatens to kill progress on generative AIs)

"When AI eats its own product, it gets sick." (FreeThink)

The images introduce how the AI. That surrounds the same data and causes corruption in the data's entirety. When data travels in wires, there is always some kind of disturbance. Those electromagnetic and physical anomalies cause a lost bit or two in the data that travels in the wire. The loss of one or two bits doesn't mean much. When the data surrounds the system thousands of times. That causes a very bad change in the data flow. And sooner or later, the data mass is corrupted and turned into another one. 

The corruption in data is the same thing as the model collapse in the data flow. 

When we use large language models LLM's we use computer programs that can search and connect data by using certain algorithms. That algorithm searches for details from search and then connects data from sources that fill certain rules. That makes it possible for the LLM can make almost suitable academic texts and translate homepages. The problem with the LLM is that it doesn't think. 

The LLM sees letters as ASCII codes and searches for similarities in the ASCII codes that it sees in homepages. So the LLM doesn't know the meaning of the words that it uses to generate new texts. The LLM collects words that look like the same ones that are in the search field. And then the AI connects those words into the new entirety. 

In that model, the LLM starts to surround the data. That means the LLM recycles texts that it made itself. And that thing forms the corruption in the code, that the LLM recycles. When LLM starts to use its texts and images as sources, that can cause problems with data, that the LLM uses. 

One of the biggest problems is that the AI can start to recycle its texts. Or the AI can interconnect the data, that other AI's generate. And when the AI generates data, it just connects sources to the new entireties. 

The AI doesn't know what words mean. And that makes it easy to manipulate its answers changing the involvements of those homepages. 

The problem is this. When the AI selects sources, it uses the homepages from the search engine. The search engine doesn't make a difference does the click come from the LLM or the human? And the LLM can raise some homepages pagerank. It's possible. Some homepages involve cool-looking information but having some kind of scientific-looking heading or title in the search path can raise the homepage that has nothing to do with the thing, that maybe the student searches. 

And that makes the AI give text, there is nothing to do with studies. In the worst case, the person who does not know the searched topics puts that AI-created crab on Wikipedia or something like that. In the worst case, the AI starts to recycle that data. And in some other cases, the AI starts to connect data, that other AI's created. In those cases the data that AI connects and rewrites starts to corrupt. 

https://www.freethink.com/robots-ai/model-collapse-synthetic-data

Comments

Post a Comment

Popular posts from this blog

What is the difference between TR-3A and TR-3B? And are those planes real?

What is the difference between TR-3A and TR-3B? And are those planes real? Is TR-3B (0) "Black Triangle UFO" or is it only the piece of paper?  The study project, what is used to create advanced ideas for use of the nuclear-powered aircraft. Or is it the study project or black budget aircraft, where lost 2,3 trillion dollars (1)of the stealth bomber were gone. In this text is things, that might seem very difficult to accept, and when we are thinking about things like doubling the object or making it smaller by using huge layers of energy, nothing denies to test those things. But were those tests successful, there is no data about that in public Internet, so we must say that things like doubling the human or aircraft can be tested, but the results can be unknown.  But in the source two is the tale, what seems like impossible, those men, who got Noble Prize put at first time one atom to the box, and hit it with photon one photon in the box and hit it with th
Post a Comment
Read more

Mythic Planet X is an interesting thing because that allows the way to the perfect stealth technology.

  Mythic Planet X is an interesting thing because that allows the way to the perfect stealth technology.  Could a hypothetical Planet X tunnel all radiation?  The myth of Planet X still lives. Nobody ever saw that planet, and the thing that supports the theory of previously unknown large-size planets in Kuiper's belt are mysterious anomalies in Neptune's trajectory. The thing is that nothing else supports the theory of the existence of the "invisible planet". The size of the planet must be very big. And its gravitational field strong that it can affect Neptune's trajectory.  And the search for that mysterious planet is very long work. In some visions, the mysterious gravitational source is mentioned as the small black hole, but nobody understands why that black hole doesn't seem to interact.  In some other theories, Planet X is a glimpse of dark matter. And that thing makes it invisible. But there is one weakness. The dark matter object must have some kind of
Post a Comment
Read more

Nanoparticles are excellent tools for medicine transporters.

"Researchers have developed a new therapy for pancreatic cancer involving nanoparticles that stimulate immune responses and improve drug delivery. This innovative method has led to significant tumor reduction in mice and holds potential for treating other cancers. Credit: SciTechDaily.com" (ScitechDaily, New Nanoparticle Cancer Treatment Successfully Shrinks and Eliminates Pancreatic Tumors) Nanoparticles can carry medicines into wanted cells. The idea is that the same system that feeds those cells transports those nanoparticles into cells that need medicine. When a nanoparticle goes, into a cell, it releases the chemicals into the targeted cells. Because nanoparticles don't let medical molecules interact with a body that makes it possible to create new and more powerful cytostatics. Nanoparticles can transport things like ricin molecules into wanted cells.  Ricin is one of the most poisonous chemicals. That chemical is useful for next-generation cytostatics, if researche
Post a Comment
Read more