From facial recognition in smartphones, to digital voice assistants like Siri, to tools like ChatGPT, artificial intelligence and machine learning are part of our daily lives. While its benefits are numerous, its rapid adoption has also raised questions about its risks. The ethical use of artificial intelligence and machine learning (AI/ML) in scientific research will also become a more visible and important consideration, according to Guido Cervone, professor of geography, meteorology, and atmospheric sciences at Pennsylvania State University. It is said that there is
Cervone contributed to a report published by the American Geophysical Union (AGU) on Principles and Best Practices for the Ethical Use of AI/ML in Earth and Space Science Research. Mr. Selborne, chairman of the AGU’s Natural Hazards Section, is one of six members of the Steering Committee and one of the report’s 20 authors.
AI/ML tools and techniques are enabling advances in our understanding of the Earth and its systems at all scales, informing key decisions by researchers, organizations, and government agencies. According to Cervone, AGU’s report, “His Ethical and Responsible Use of AI/ML in Earth, Space, and Environmental Sciences,” is designed to support these advances while mitigating potential risks. increase.
“We are collecting more data than ever about all aspects of the universe, from the inner core of the Earth to stars far outside our solar system, and we are using computational approaches to bring this data together. We’re doing more analysis,” said Brooks Hanson, AGU’s executive vice president of science. “This is an incredibly exciting time for science, but rapid changes like this can create ambiguity in how scientists do their research. We need to ensure that the limitless possibilities offered by are balanced by clear ethical standards, enabling researchers to conduct research responsibly in ways that benefit the larger scientific community. .”
Ethical use of AI/ML in research requires a new way of thinking about methodologies, according to the AGU. For example, validation and replication are core principles of science, but in AI/ML-powered research this can be complicated as the inner workings of models can be opaque. Traditionally, research had to describe the entire scientific process in detail, but AI/ML-powered research can document only the steps of the process, not the actual calculations that result. Additionally, AI/ML-powered research should document potential biases, risks, and harms, especially those related to promoting justice and fairness.
“Trust is an important theme for AI/ML research, but it’s not the one we’re trying to answer today,” said Thurborn, who is also deputy director of the Computational Data Sciences Institute at Penn State University. “Today’s AI/ML methods often represent knowledge in forms that are difficult to validate and understand, and thus lack some mechanisms for evaluating the reliability of results. It requires a degree of confidence that is not commonly discussed with other analytical techniques historically used in the Earth Sciences, and that there are many opinions in this area that we need to continue discussing this in detail. is clear.”
AI/ML is a powerful tool for evaluating diverse datasets. It helps Earth, space, and environmental scientists uncover new insights about our planet, warn communities about natural disasters such as tornadoes and wildfires, and improve scientific predictions, including climate-related predictions for the future. It helps you to Risks such as sea level rise.
“When it comes to determining biases and uncertainties in datasets and models, researchers are increasingly improving how they document these details to make them available,” said AGU’s Open Science Leadership Officer. Vice President Sherry Stoll said. “Correspondingly, there has also been a significant increase in Earth, space, and environmental science research utilizing AI/ML methods. It provides ethical guidelines to inform researchers and their organizations of the importance of linking them to decision-making about composition and workflow.”
Funding for this effort was provided by NASA.
