Is AI the missing link connecting our past and future?

In a world preoccupied with forward progress, can examining human behavior in deep time actually inform our collective future? According to Michael Barton, professor in ASU's School of Human Evolution and Social Change, as well as a professor in the College of Global Future’s School of Complex Adaptive Systems, the answer is “yes”—and artificial intelligence (AI) can help.

In a recently released commentary, "Collaboration between artificial intelligence and Earth science communities for mutual benefit," where Barton was a coauthor, it was argued that AI could be the bridge between understanding commonalities and patterns across complex data sets.

“I think this is a forward looking commentary. Currently, and in a lot of Earth science research, complex data-driven models are not used a lot. Places where Earth science is already using AI are things like pattern recognition from satellites and remote sensing like Google Earth. But, Earth scientists, just like archaeologists, generate huge amounts of complex data. That data is useful for looking at questions that can be addressed through these kinds of machine learning,” says Barton.

The authors suggest that generative AI tools could be used to distill down complex data sets and predict future outcomes. Whereas predictive modeling relies solely on historical data, the authors say the learning capabilities of AI could outperform traditional models in terms of accuracy.

"We have to be careful about trying to predict the future from the past in just linear ways. We need to involve other kinds of approaches beyond data-driven AI models, such as process models, as a way to do that," says Barton.

However, using AI to analyze complex data sets is only one part of the equation. According to Barton, the other half of the equation may lie with understanding the connection between human risk and resilience.

Barton and four coauthors published "Lessons for an invisible future from an invisible past: Risk and resilience in deep time" in The Holocene, where they suggest that while one can't draw a direct comparison between humans in deep time—thousands of years ago—and the modern world, risk and resilience are relevant fundamental principles shared by both.

"We’re trying to make the case that you can't just draw a straight line between what people did in the past and what they're doing today or in the future. You have to try and somehow go below that and extract … fundamental principles that are relevant, in this case, to risk and resilience," says Barton. "If you can do that, then maybe you can actually apply some of what we've learned about risk and resilience in deep time to things today—even if you can't make any kind of a direct comparison."

Resilience, as used in the archaeological context, refers to humans' ability to manage risk through cultural knowledge, social practices and technology. However, both in the past and present, resilience can also make people vulnerable to additional unforeseen risks, Barton says.

One example the paper gives is the land use strategies of human ancestors thousands of years ago in Europe. Periods of global cooling and warming created environmental uncertainty and led to resource shortfalls. As an adaptation, Barton says that ancient human hunter-gatherers began to employ logistical land use and mobility to contend with greater risk in access to food, shelter and water. Logistical land use is when hunter-gatherers set up a "base camp" in a favorable location and then send out foraging parties to bring back food and other resources to camp.

"Because logistical land use involves regular long-distance forays to locate patchy, mobile animal resources, it means that logistically organized human groups have an increased opportunity to meet and interact with a wider community of more distant human groups than they would following residential land use strategies," Barton and his colleagues write.

Through the act of finding resilience to scarce resources, prehistoric humans exchanged cultural knowledge and genes with a diversity of other human groups. As Barton and his colleagues point out, the unforeseen risk was ultimately a loss of distinctive biological and cultural traits—a kind of globalization that led to the disappearance of the original Eurasian population.

"Neanderthal descendants with some Neanderthal traits survived, but Neanderthals as [a] recognizable population with distinctive biological and cultural traits did not," the authors write.

For a modern example of this kind of resilience and globalization risk, Barton says he has to look no further than his research base in Spain. In the 40 years since he began researching in the country, he has witnessed a major shift in the society.

"When I first started working there, it was shortly after the end of the Franco era, and so I can see the many benefits that have happened. But, at the same time, I've commented to some of my Spanish colleagues that if I go to some of the cities now, they look a lot like any other European city with similar architecture, similar kinds of dress, similar kinds of activities. There are differences still—but there is this globalization that has gone on," Barton says.

"As we try to point out in the paper, a lot of people talk about resilience, but it is something that is actually very hard to measure, especially because it inherently refers to the future," says Barton.

While AI may be the tool that helps Earth scientists and archaeologists use the past to predict the future, Barton says it isn’t a crystal ball.

“There's never going to be an ideal solution, there's only going to be a better solution,” says Barton. “The best thing to do is find out how you can be most broadly resilient and maybe in different dimensions, maybe to try and protect against some risks, but be flexible enough that you can absorb and bounce back from other kinds of risks, even if you can't protect against them.”