A group of researchers have published a new study indicating the possibility of creating smaller models of artificial intelligence without human intervention.
Large artificial intelligence models like the ChatGPT robot model can build more customized and compact applications for everyday use.
This issue was discussed through collaboration between Aizip Inc and scientists from the Massachusetts Institute of Technology and several universities in California.
These specialized models can perform various tasks such as improving listening devices, monitoring oil pipelines, and others.
In this context, Yan Sun, CEO of the artificial intelligence company, stated: “We are currently using larger models to build smaller models, resembling the cooperation of an older sibling in assisting the younger one. Undoubtedly, this is the first step towards achieving a greater goal of building self-evolving artificial intelligence.”
Yuobi Chin, a university professor and co-founder of Aizip, highlighted the remarkable discovery that the larger model can contribute to designing smaller artificial intelligence models automatically. Therefore, he believes that these large and small models can collaborate and build an intelligent ecosystem.
Chin stated: “This technology represents significant progress; for the first time, we can design a fully automated pipeline. The model can design a smaller model of artificial intelligence without any human intervention.”
We presented this month the first evidence of this concept, showing the possibility of designing one of the models from data creation to deployment without human intervention.
A device was reviewed by Sun and Chin, a device embedded in a small electronic chip no larger than a dime. This device uses artificial intelligence sensor to track user activity, collect and analyze motion data.
This sensor is a clear example of small machine learning models and small artificial intelligence systems that can be used in devices. There is no doubt that machine learning capabilities are the foundation of the spread and essence of artificial intelligence, essential for the idea of transforming all things into smart.
Chin commented: “If we look at the ChatGPT robot and the small machine learning model, we will find that they are at opposite ends in terms of artificial intelligence. The large models are present in the cloud, while the small models are present in objects.”
The research indicates the possibility of incorporating artificial intelligence in the future into many daily household devices such as coffee makers, ovens, and televisions.
