Site icon Stories Cover

SHI DALL-E 2 has created its own language that cannot be deciphered

Various artificial intelligence systems are already capable of creating things that look completely alien to humans. Moreover, there have been cases in history where artificial intelligence systems have invented their own language, once it alarmed researchers so much that they even preferred to reset the system, breaking the chain of intelligence that went in a direction incomprehensible to them. This time, the DALL-E 2 system, created by Google specialists, was noted, which was originally intended to generate images based on a textual verbal description.

Note that the neural networks of the DALL-E 2 system were pre-trained on a huge number of images, accompanied by a detailed text description. Built neural connections form a kind of database in which the relationships between symbols, words and image elements appear.

“Miracles” with the DALL-E 2 system were noticed when it was given the task of independently generating a text description and the corresponding image. It turned out that the system in its bowels operates in its own incomprehensible language with its own words for the vast majority of nouns.

The researchers believe that the main reason for what happened is that the description of the images on which the system was trained, were presented in different languages. This forced the system to bring these descriptions to some “average denominator” and compare the result with the elements of the images. The principles used by the system to create the “middle denominator” are still a mystery to scientists, as is the mystery of a kind of new language consisting of averaged data.

Researchers see a potential danger in the fact that the artificial intelligence system creates its own language. Imagine, if this language is deciphered by a person who has not very good goals, then this person will be able to make this language a task for artificial intelligence, which, performing it, will attack the Internet or other critical infrastructure. And other people will not even be able to understand what is happening and for what reason.

That is why researchers are now keeping the DALL-E 2 system isolated from the rest of the world, observing the processes taking place in it, and then what will lead to its further development. However, it is worth mentioning that in the world there is another version of this system, DALL-E Mini, public, but has much more limited capabilities compared to the DALL-E 2 system.

Exit mobile version