The fearsome AI has seen the creation of its own science that human experts don’t understand

The fearsome AI has left scientists baffled after discovering a physics that professionals still don’t understand.

Physics is one of the most rigorous scientific disciplines with complex equations and precise measurements to reveal secrets.

4

4

It’s a task that doesn’t have a good path to follow, said Columbia engineering professor Hood Lipson.

“It’s an art, there’s no systematic way,” he told Motherboard.

“It is almost, how do you discover the alphabet? It happens naturally.”

In the Creative Machines Lab, Lipson and colleagues aim to better understand how this discovery process occurs and how it can be improved using machine learning.

The team developed an algorithm capable of studying physical phenomena by “watching” videos showing the flash of flames or the swing of a double pendulum.

The algorithm was able to predict the correct number of variables within known systems and even make predictions for unknown variables.

The team published their findings last week in a study titled Automated Detection of Key Variables Hidden in Experimental Data in the journal Nature Computational Science.

Lipson said that this work differs from other attempts to study similar data because it is the first to not provide the algorithm with any information about the number or variables in the system.

This means the system isn’t just looking for variables with a human gaze, which Lipson said could be crucial to finding the physics hidden within systems.

“It’s not that people toil day and night looking for these variables and that can speed up the process,” Lipson said.

“It’s more that we’re probably ignoring a lot of things, but a lot depends on those variables that we thought if we could use some of the power of AI on this maybe we’d discover things that are very useful and will change the way we think about it.”

The team, including the first author of the research paper and an assistant professor of engineering at Duke Poyuan Chen, provided videos of the algorithm showing the dynamic motion.

The videos also included still-unintelligible movements such as lava lamps and inflatable air dancers.

AI attempted to model this phenomenon after studying the videos to create a list of increasingly smaller variables.

Then, it will give the minimum number of variables the system needs to accurately capture the movement.

AI has succeeded in figuring out the right number of variables but it currently lacks the language to describe what the variables are.

This will prevent him from entering science labs anytime soon, but Chen thinks it’s not a big deal for now.

“What we have now is a general framework,” said Chen.

“The only thing that would be interesting is collaborating with experts who have data and intuition about what that data does. What we want to do is help them discover what they don’t yet know about the data.”

Lipson believes that the algorithm could study systems beyond physics, such as disease evolution or future climate change.

The team hopes that the algorithm will help communicate its findings more easily to humans, as it could potentially be a major advance in scientific discovery.

“Humans have been doing this for 300 years, and it seems to me that we are coming to the end of what we can do manually,” Lipson said.

“We need something to help us take it to the next level.”

4

4