Computer-based models have become an essential part of science, technology and even decision-making.

Hydrology and atmospheric sciences professor Hoshin V. Gupta says any model is “a simplified representation of reality.”

Models can be used to make predictions about how the world is going to behave and to enable people to come up with engineering solutions, but importantly we use models to help improve our understanding of the nature of the world/universe, which is how science proceeds, he says.

In Gupta’s area of expertise, models can be employed to forecast floods or make any number of predictions about water availability and quality.

“The same principles that govern hydrologic systems also govern other kinds of systems in the world,” Gupta says.

Gupta says the human species has throughout its evolution become particularly good at making individual (mental) and group (collective) models that help us to “ensure our survival and navigate the world.”

“But if you go to a person and say ‘Why, using all that you know, did you make this decision?’ they might be hard-pressed to explain exactly why — the nature of their own informal mental models may not be clear to them.

“Scientists try to build formal conceptual, mathematical and computational models to help us understand, explain and predict the behavior of the world so we can optimize our decisions.”

Humans make models individually (in our heads); when that’s done communally as groups it can lead to better and more explicit models, and ultimately to the process we call science, he says.

“In science, we like to focus on being able to explain why an answer is the correct answer,” Gupta explains.

“Historically, science has proceeded by summarizing what we could learn from experiences, observations and data into laws and principles that explain how the universe works. Scientists then use those principles to build computational models to try to make predictions about the world,” he says.

Recently, however, things have begun to change. “Now there is this emerging ability to take huge amounts of data and put it into a computer and to search for patterns in that data that can be used to make predictions, something that human brains find difficult to do. However, computers are not necessarily able to provide insights into why those patterns exist.”

While machines are getting better at analyzing data, Gupta says human beings somehow have the ability to generalize from these analyses and provide insight “in a way that we have not necessarily been able to get machines to do.”

This, of course, poses an important challenge: “How do we exploit the powerful computational and data processing abilities of machines, while getting them to emulate and extend upon human analytical abilities, thereby greatly enhancing our understanding of the world?”