By Tim Stobierski
May 2024

It might seem obvious today that artificial intelligence (AI) could be a helpful tool in discovering polymers for materials development. After all, the technology has been applied to challenges as diverse as drug discovery, protein synthesis, and more.
But in 2020, when Ares Materials first made the decision to build an AI program to aid their materials discovery and product development cycles, there wasn’t a lot of certainty as to whether or not it would work.

“It wasn’t as clear a path then that you could actually leverage AI for materials development,” says David Arreaga, the company’s CEO and Co-Founder.

The challenge? Training AI models requires a lot of data, and there weren’t enough large datasets available to enable this training.
Ares Materials believed that they could get around this challenge by translating their human knowledge of chemistry and polymer science into algorithms capable of making generalizations from the small datasets that were available.

“At the time, putting this strategy in place was a risk,” says David, “because we couldn’t be sure that it would work. But in the past six months, we’ve seen it begin paying off.”

Traditionally, you’d test one variable at a time to determine if a material would work for the purpose you intended. With Ares’ AI model, it’s possible to test many variables at once, decreasing the number of experiments needed to advance from discovery to product development.

“It’s been astonishing how much AI has reduced our need for experimentation,” says David. “In the past it would take us six months to reach the same conclusions we can now reach in a month.”

Ares is already in advanced mass production qualification for one new substrate identified by their AI program, and they are currently developing five additional materials for other cutting-edge applications. David expects that the next 12 months will be an exciting time, as the program takes off in earnest and the efficiencies the company realizes from it begin to compound.