Large AI systems can “give birth” to small subsystems

New research published by a team of scientists from the Massachusetts Institute of Technology and the University of California reveals the unusual abilities of artificial intelligence.

As Fox News reports, large artificial intelligence models are now capable of creating their own subsystems without human intervention.

These results apply to large neural networks similar to those used in modern models, including ones like ChatGPT. These neural networks are now capable of not only performing complex tasks, but also creating smaller but highly specialized artificial intelligence applications.

The application of these subsystems is varied and covers various areas. From improving devices like hearing aids to monitoring oil pipelines and monitoring endangered species, these small but powerful apps promise to transform our daily lives.

Scientists emphasized that this step in the development of artificial intelligence is like an older brother helping a younger brother in his development.

This is just the first step towards creating larger-scale systems of self-developing artificial intelligence.

However, what makes this discovery truly amazing is not only the ability to create smaller models without human intervention, but also the ability to use parent algorithms to automatically generate these models.

Scientists foresee that in the future, large and small AI models will collaborate, creating a complete intelligent ecosystem.