Artificial intelligence can learn from babies about the physical world

A new artificial intelligence (AI) system can interpret physics and how the physical world operates. The system has done this by being modeled to mirror how babies react to changes in their physical world. This system has been hailed as progress in the field of machine learning.

A team of MIT researchers has been to develop an AI that can learn and understand how basics work. The model is able to understand some basic principals of physics as some humans. By humans, we are referring to three-month-old babies.

Three-Month-Old babies already have some concept of how the world around them works. They are able to understand advanced physical concepts such as solidarity and permanence. They understand that objects do not just appear and reappear out of thin air. There has to be some progression of motion.

Artificial intelligence interprets the physical world

This is well documented where researchers were able to put motion pictures in front of babies. They used objects passing behind a wall. They observed that the reaction was normal when the objects reappeared on the other side after some time. However, when it did not appear or did not follow the common physics laws, the babies acted very surprisedly, proving they understood what was going on.

MIT researcher Kevin Smith said that these babies understood that objects do not wink in and out of existence. He explained that this was very important in their decision to create artificial intelligence that could be able to copy the brain of three-month-old babies.

He continued by saying,

We wanted to capture and formalize that knowledge to build infant cognition into artificial-intelligence agents. We’re now getting near human-like in the way models can pick apart basic implausible or plausible scenes.

Anomalies in simulation movements

The researchers, therefore, embarked on a mission to create an AI that could determine if a physical phenomenon was surprising or not. To measure this, the researchers used a computer-aided simulation of physical events. The AI would then try to determine if there are any anomalies in the way these simulation movements are behaving. They would later make an informed decision on whether it followed the laws of physics or not.

This research is helping unravel the mystery of the human mind while still trying to advance AI technology. In the future, researchers may be able to develop more advanced artificial intelligence systems that will be able to think for themselves, something that will be a great scientific milestone.



Featured image by Pixabay