The burgeoning space industry is increasingly embracing artificial intelligence (AI), utilizing technologies ranging from large language models to neural networks to solve complex challenges. Orbital Sidekick, for instance, employs a ChatGPT bot to monitor data streams, as COO and co-founder Tushar Prabhakar explained at the SmallSat Symposium: “one of the hardest things for us was keeping track of where and when the datasets for customers were coming through…We can ask [the bot] questions and it alerts us when a dataset is here.” Future applications include leveraging these models for historical data analysis to pinpoint previous events like oil pipeline leaks.
Orbital Sidekick is also utilizing neural networks to address the issue of false alarms in geospatial data. Prabhakar noted, “Not many people talk about false alarms in the geospatial world…When you’re detecting methane or liquid hydrocarbon, they’re everywhere.” This highlights the practical, often “unglamourous,” applications of AI, as emphasized by Alan Campbell, AWS principal space solutions architect: “Summarizing documents, improving code efficiency, these are great uses for large language models.”
While AI won't replace human workers, as Millennium Space CEO Tony Gingiss points out, it can significantly improve efficiency. He observes, “A lot of people struggle to keep track of the hundreds of emails they receive daily…Can we figure out how to process that data and get through the noise to the important stuff? That will be a transformative thing.”
Innovative AI applications are emerging across various sectors. Iceye, a Finland-based company, uses generative AI to pair satellite imagery with social media data for flood level assessments, creating “first-pass situational reports to local authorities,” according to Campbell. Similarly, Degas Ltd., a Tokyo-based firm, developed a chatbot providing targeted agricultural advice to farmers in Ghana based on geospatial data.
However, caution is warranted. Lucy Hoag, founder and CEO of Violet Labs, shared a recent experience where a Microsoft Copilot-suggested recursion function, while initially efficient, caused significant performance issues at scale: “If you grow out of that scale, it becomes really terrible…It’s a double-edged sword.”