How Ecology Could Encourage Higher Artificial Intelligence, and Vice Versa
A lot of today’s artificial intelligence systems loosely mimic the human brain. In a latest paper, researchers suggest that one other branch of biology—ecology—could encourage an entire latest generation of AI to be more powerful, resilient and socially responsible.
Just published in Proceedings of the National Academy of Sciences, the brand new paper argues for a synergy between AI and ecology that might each strengthen AI and help to unravel complex global ecosystem challenges, reminiscent of disease outbreaks, lack of biodiversity and the impacts of climate change.
“The issues we face are so urgent, we are usually not developing theories on the speed we’d like to, to deal with them. AI can has the potential to assist us jump from broad data to actual knowledge without going through all the standard steps,” said study coauthor Ajit Subramaniam, a biological oceanographer at Columbia University’s Lamont-Doherty Earth Observatory. Conversely, he said, “The really exciting thing is that AI not only helps us as scientists in numerous fields, however the principles of ecology also can help AI move forward.”
A picture generated by the AI system DALL-E using the prompt, “a synergistic future for artificial intelligence and complicated ecological systems.” Courtesy of Barbara Han/Cary Institute
The concept arose from the authors’ statement that AI could be shockingly good at certain tasks, but still removed from useful at others—and that AI development is hitting partitions that ecological principles could help it overcome.
“The sorts of problems that we take care of repeatedly in ecology are usually not only challenges that AI may benefit from when it comes to pure innovation; they’re also the sorts of problems where if AI could help, it could mean a lot for the worldwide good,” said Barbara Han, a disease ecologist at Cary Institute of Ecosystem Studies, who co-led the paper. “It could really profit humankind.”
Ecologists, Han included, are already using artificial intelligence to go looking for patterns in large data sets and to make more accurate predictions, reminiscent of whether newly discovered viruses may be able to infecting humans, and which animals are almost certainly to harbor those viruses. Nonetheless, the brand new paper argues that there are various more possibilities for applying AI in ecology, reminiscent of in synthesizing big data and finding missing links in complex systems.
Scientists typically try to grasp the world by comparing two variables at a time. For instance, how does population density affect the variety of cases of an infectious disease? The issue is that, like most complex ecological systems, predicting disease transmission will depend on many variables. Ecologists don’t at all times know what all of those variables are; also they are often limited to physical ones that could be easily measured, versus social and cultural aspects.
“In comparison with other statistical models, AI can incorporate greater amounts of knowledge and a diversity of knowledge sources, and that may help us discover latest interactions and drivers that we may not have thought were vital,” said study coauthor Shannon LaDeau, a disease ecologist on the Cary Institute. In helping to uncover these complex relationships and emergent properties, artificial intelligence could generate unique hypotheses to check and open up latest lines of research, she said.
Artificial intelligence systems are notoriously fragile, and once they go awry, there could be potentially devastating consequences, reminiscent of automotive crashes (already a reality with self-driving cars) or a misdiagnosis of cancer (if medicine becomes sufficiently depending on AI).
The incredible resilience of ecological systems could encourage more robust and adaptable AI architectures, the authors argue. Specifically, ecological knowledge could help to unravel the issue of so-called “mode collapse” in artificial neural networks, the systems that power speech recognition, computer vision and other functions.
“Mode collapse is once you’re training a synthetic neural network on something, and you then train it on something else, and it forgets the very first thing that it was trained on,” explained Kush Varshney of IBM Research, who co-led the paper. “By higher understanding why mode collapse does or doesn’t occur in natural systems, we may learn easy methods to make it not occur in AI.”
Inspired by ecological systems, a more robust AI might include more flexible feedback loops, redundant pathways and decision-making frameworks. These upgrades could also contribute to a more so-called “general intelligence” for AI systems that might enable reasoning and connection-making beyond the particular data that the algorithm was trained on.
Ecology could also help reveal why AI-driven large language models, which power popular chatbots reminiscent of ChatGPT, show emergent behaviors that are usually not present in smaller language models. These behaviors include “hallucinations”—when an AI system generates false information. Because ecology examines complex systems at multiple levels and in holistic ways, it is nice at capturing emergent properties reminiscent of these, and will help to disclose the mechanisms behind such behaviors, the authors say.
The paper was also coauthored by Kathleen Weathers of the Cary Institute and Jacob Zwart of the U.S. Geological Survey.
Adapted from a press release by the Cary Institute.
