The state of AI in today’s geospatial industry

The ongoing AI revolution, marked by rapid advances towards artificial general intelligence (AGI), is transforming the way we live and work. This is already having an impact in programming and general data management jobs, but how is this affecting the geospatial data scene? This article highlights the latest advancements of AI in this industry, and informs geospatial specialists of what to expect in the coming year and how to love, rather than fear, AI.

Artificial intelligence (AI) is growing at an exponential rate. From assistive tools to autonomous agents. AI systems are no longer limited to answering questions; they can now plan, execute and evaluate multi-step tasks with minimal human oversight. Agentic AI can operate software, coordinate workflows and make decisions based on context, changing how humans interact with digital systems.

This growth is nothing short of explosive, and control of these systems is deemed by many to be a true necessity rather than an annoyance. Therefore, governing bodies are constantly trying to control and regulate the growth and implementation of AI for public safety. The EU AI Act is establishing global standards for transparency, risk management and accountability. As it comes into force, organizations deploying AI in Europe must demonstrate compliance and explainability. This will shape how AI systems are designed and deployed worldwide.

In the meantime, the growth of AI is largely limited by its implementation. This means the technology and processing power are there, but what’s lacking is our ability to provide the right information, and ask the right questions. For example, the video game industry has always been a benchmark for digital technology, and AI is no exception. What started with AI text generation soon progressed to images and video, and AI is now able to create entire video games from a single prompt. These systems can dynamically generate environments, characters, narratives and gameplay logic, which is positioning AI as a general-purpose creative engine rather than a set of narrow tools. Needless to say, this does not only count for video games, but for all types of computer programs that can be produced, launched and maintained by AI.

From tool-driven to intent-driven GIS workflows

The geospatial sector is feeling the effects of these changes and innovations just like the rest of the world, and in some cases it is being hit even harder. AI is reshaping the geospatial sector by shifting GIS workflows from tool-driven to intent-driven. Instead of manually chaining software tools and scripts, users increasingly describe what they want to know, while AI systems determine the data sources, processing steps and validation automatically.

Tools like Esri’s ArcGIS GeoAI toolbox are at the forefront of this automation wave in the geospatial sector, incorporating machine learning for classification, regression and natural language processing on diverse datasets. Esri’s GeoAI helps users to automatically classify and transform complex point clouds and datasets into usable objects. In a recent interview with GIM International, Esri’s CEO Jack Dangermond said: “Much like scanning a barcode, you can update a city’s vegetation inventory or a utility company’s asset condition data almost instantly.” He was not only referring to once-a-year tasks, but also to more repetitive tasks that take valuable time out of each day for employees in the public and private sector. Over the past decades, Esri has built up a strong reputation for accurately predicting the way the industry is heading. Based on the company’s views on AI, these tools will cause a substantial shift.

The industry is also transitioning from batch analysis to near-real-time and streaming Earth observation. AI pipelines increasingly ingest continuous satellite data streams, allowing automated processing as new imagery arrives. As a result, change detection alerts can be generated within hours instead of weeks, supporting faster responses for disaster management, infrastructure monitoring and environmental compliance.

Radar and optical data fusion

One major technical advance is the move towards radar and optical data fusion. AI models now routinely combine synthetic aperture radar (SAR) data, such as the aperture of the Teide volcano in Figure 4, with optical imagery from Sentinel-2 or commercial satellites. This enables reliable monitoring regardless of clouds, darkness or seasonal conditions, making year-round observation practical even in challenging environments.

The latest, and perhaps most crucial, shift for the geospatial industry is AI’s recent ability to work with vectors and point clouds, where before this was exclusively rasters. By working with these more inclusive, scalable and complex data formats, AI systems can work directly in projects for supporting roles with digital twins, urban analytics and infrastructure planning. Together, these developments are transforming geospatial data into continuous, actionable spatial intelligence – not only for geospatial specialists themselves, but also in terms of how they work together and communicate with different domains.

The digital-to-physical link

At its core, AI operates entirely within the digital domain; it can process, generate and analyse any information that can be represented as data. This makes it extraordinarily powerful in text, images, code and structured datasets that exist in digital form. However, the physical world remains outside AI’s direct reach. It cannot sense or control reality without a digital interface translating physical phenomena into data it can interpret.

This physical-to-digital ‘translation’ layer is where spatial data plays a critical role. Spatial data captures the geometry, location and attributes of the real world, turning physical entities into digital representations. That being said, spatial data by itself is not enough. AI needs context: scales, rotations, translations, transformation and interpretations – all crucial information to make sense of generic spatial data. This involves the gathering of different data sources, interpretation of these sources with context for each source, and informed decision-making.

Geospatial specialists are irreplaceable

This is where spatial data specialists are irreplaceable. They receive data and context, decide which data is required if none is available, and bridge the gap between the digital and the physical world. Specialists are needed for every step in data processing, from collection to validation and publication. Whether strategizing model placement, validating model outputs, ensuring data quality and provenance or interpreting spatial patterns within real-world contexts, a controlling party with full context information is necessary.

What’s next?

Some people say that AI, with all its hype, is just another bubble – just like the internet, it will not deliver on its value promise. Whether this is true or not, here are a few things that are likely to be on the horizon in the coming year:

1. Reliable autonomous AI agents
   AI agents will move from experimental to operational use, handling complex, multi-step tasks end-to-end with built-in self-checking and escalation.
2. Real-time, multimodal AI systems
   AI systems will increasingly operate in real time while natively combining text, vision, audio, video and sensor data. This is essentially a crucial step for the true agentic AI systems crossing into our world.
3. AI designed for regulation and trust by default
   With regulations like the EU AI Act coming into force, AI systems will be developed with built-in compliance features: traceability, explainability, confidence scoring and audit logs. Even though such regulations slow AI growth, we rely on the EU and other governments to maintain a grip and ensure the safe development and usage of AI.

Preparing professionals for the transformation

So how should today’s professionals prepare themselves for this transformation? As Esri’s CEO Jack Dangermond noted, it is impossible to predict exactly what the next decade of technological advancement will bring. What is clear, however, is that repetitive tasks are rapidly disappearing. Work in the geospatial sector – both in the field and during desktop processing – is increasingly focused on determining what is needed, rather than on executing every step required to obtain it.

“Our industry is moving beyond the hype of AI faster than others because we have so many repetitive tasks it can improve,” said Ron Bisio, Trimble’s senior vice president of field systems, in a recent interview with GIM International. Surveyors know this better than anyone, and understand that thinking on your feet and solving problems as they arise is key to being a good surveyor. As automation is implemented at an accelerating pace, understanding both the capabilities and the boundaries of AI systems will become a core professional skill. When facing a challenging setup with a laser scanner, for example, field professionals should ask themselves how AI could perform that task, and where it would encounter limitations.

The same applies to data processing workflows. Any task that is performed repeatedly is a strong candidate for delegation to AI assistants or autonomous agents. Data specialists must remain critical and reflect on the obstacles they encounter during processing, and on the decisions that fall outside standard workflows. How was a specific result produced, and which steps were required beyond the default processing pipeline? Such considerations will remain essential, as these workflows increasingly serve as the foundation on which AI-driven processing systems are trained and refined. Having a good understanding of where AI may take place in their workflows the coming years will help data specialists shift their focus and adapt this fast new helper in their own work.

The educational shift

Looking to the next generation of talent, how can educational institutions prepare tomorrow’s land surveyors and geomatics specialists to work effectively alongside AI, especially since our sector is one of the early adopters? Today, many university curriculums are task-focused and tool-focused in order to give students a sufficiently profound understanding of the discipline’s various topics. However, as AI will take over many of the tasks these subjects cover, it is becoming more relevant to train the new generation in intent-driven decision-making. As the ‘what’ and the ‘how’ are disappearing, students should be encouraged to think about the ‘why’. In other words, to protect the future of the industry, schools and universities must move beyond teaching software operation and procedural workflows alone, and instead emphasize critical thinking, system understanding and AI literacy.

Conclusion

As artificial intelligence continues to advance and robotization increases, the gap between the digital and physical worlds is steadily decreasing. Autonomous surveying robots will increasingly enable data collection, evaluation, processing and delivery with minimal human intervention. Does this development imply a complete AI takeover of the profession? No. Much like ChatGPT, which initially caused uncertainty but has since become a common household tool, AI is evolving into a practical assistant rather than a replacement. Geospatial specialists should embrace these technologies and benefit from the increased efficiency, quality and volume of work they enable. This technological shift is not temporary; it is here to stay.