In this pursuit, researchers have just unveiled a significant new advancement in Geographic Information Systems (GIS). They released GIS Copilot, a research collaborative desktop assistant built to improve human-agent interaction. This innovative tool operates similarly to popular AI models like ChatGPT and Google Gemini, enabling users to generate geospatial data and maps with minimal human intervention. Alongside the creation of GIS Copilot was a deep case study. It demonstrates the incredible capabilities that come from combining artificial intelligence with traditional GIS workflows, transforming how we conduct geospatial analysis.
This completed study with three agents, which GIS Copilot entered in the 2023 FedEx Innovation Challenge, highlighted GIS Copilot’s abilities to automate geoprocessing workflows and conduct spatial analysis. This is a huge step in the right direction. This opens new possibilities for developing autonomous GIS systems that can proactively and intelligently address complicated geospatial problems.
The Role of GIS Copilot
GIS Copilot helps connect human users with advanced AI technology, enabling them to work together in new collaborative ways to tackle complex geospatial tasks. By taking advantage of more complex AI algorithms, it’s able to automatically generate workflows that once took a significant human effort to produce. Our researchers Zhenlong Li and Cervone were instrumental in creating this tool. They went all in on its promise to fundamentally change the GIS landscape.
“Just like the paradigm shifts of the past, autonomous GIS represents an emerging paradigm of integrating AI with GIS, where it is not just another tool but instead becomes an artificial geospatial analyst able to use GIS tools to solve geospatial problems,” – Zhenlong Li.
One of the most impressive features of GIS Copilot is its capability to produce maps with minimal human intervention. Users enter their requests in plain English, and the system understands complex commands to generate rich geospatial visualizations. This change makes the mapping process much more user-friendly. It allows practitioners to focus on more strategic-level planning vs. just being mired in day-to-day activities.
Introducing LLM-Find
LLM-Find has been central to this project. It serves as a dynamic data retrieval agent, automatically fetching critical geospatial datasets in real-time directly based on the user’s request. More evenly spaced out, LLM-Find can be equipped with data on almost any subject. This is everything from sidewalks to broader road networks, school locations, and even high-resolution remote-sensing imagery. Our researchers recently shared their original findings using LLM-Find to the International Journal of Digital Earth. This line of work highlights the potential to advance AI applications to the GI field.
“LLM-Find demonstrated that autonomous GIS agents can handle data acquisition from sources without manual dataset hunting, helping to reduce the grunt work of data preparation in spatial analyses,” – Zhenlong Li.
Though LLM-Find offers remarkable automated functions, researchers warn that human supervision is still tremendously important. Due to the current limitations of the agent, it requires attentive stewardship to maintain and improve data quality and applicability.
“But the number of sources the AI agent can consult is still limited, so human oversight and management is needed for LLM-Find,” – Zhenlong Li.
The addition of LLM-Find to GIS workflows is part of a larger, geospatial automation moment. This advancement is allowing practitioners to leverage large amounts of data at a scale and speed not previously possible.
Future Implications for Education and Workforce
While these advancements might make our world more technologically impressive, they create unique challenges and opportunities. This has been particularly problematic for education inside the GIS community itself. Zhenlong Li reiterated the need to be responsive to changes in the classroom and workforce. These innovations require a willingness to change on the part of both students and educators.
“It’s now more important for students to understand process or spatial thinking procedures, to learn how to learn in the age of AI in GIS,” – Zhenlong Li.
Li underscored the need for educators to be informed on rapidly-changing trends. Each of these trends can greatly influence how we teach GIS. As we all know, rapid change in technology has taken place this last few years. Academic institutions need to quickly change their curricula to train students for the new jobs these innovations are going to foster.
“In the last five years, we have seen more progress in GIS than I thought I was going to see in my lifetime,” – Cervone.
Development of next-generation AI-powered GIS agents creates an exciting and hopeful scenario for the state of geoinformatics in the coming years. These tools are getting more sophisticated and more commonly used. They promise to significantly increase efficiency and effectiveness in addressing multifaceted spatial queries.