LO
Get told when a verified record lands on this occupation → · Mark which of these tasks are yours (VOLO Pro, free during the launch) →
GIS analyst / cartographer
Turns imagery and survey data into maps and spatial databases: digitising buildings, roads and land cover, running spatial analysis, maintaining the data, and checking it against the ground. Machine learning is taking over bulk feature extraction at national scale — France's mapping agency says AI cut the time to produce its land-cover map to a third and halved the cost — and AI agents can now carry out basic GIS operations. But in a controlled test, experienced mappers working by hand were faster and more accurate than an AI-assisted editor, benchmarks find spatial reasoning the hardest part, and even the providers of machine-made building data tell users to check it locally. The US projects cartographer employment to grow 7 percent to 2035, without mentioning AI.
Compare a machine-made building layer with imagery for one neighbourhood you know and count the errors.
This is not a probability of losing your job. It combines how much of the role's task load is exposed to automation with how far adoption has actually gone — useful for comparing occupations on one consistent basis, and for nothing else.
Written for GIS analysts and technicians, cartographers and photogrammetrists who produce maps and spatial data. Licensed land surveyors, civil engineers and data scientists have their own pages or are different jobs. The evidence is two US labour projections, a national mapping agency's account of its own production, a technology company's open building dataset, two benchmarks of AI agents on GIS tasks and a controlled mapping experiment; it establishes what AI does in map production and where people still check it, not how GIS jobs or incomes have changed.
What is actually changing#
The unit of analysis is the task, not the job title. A role is not replaced — its task mix shifts.
Each tile is one task. Its size is how much of the job it is; its colour is where the task is heading. Click a tile to see what the judgement does not establish.
An agency's account of its own production, a vendor's claims about its own data, and one small experiment; none measures mapping jobs.
Benchmarks in test settings; valid workflows are not the same as correct answers, and neither measures analysts' work.
Projections for one country; no record measures AI in map design or database maintenance.
A projection's reasoning and a vendor's advice to its users; neither measures how much checking is done.
Is this your job? Say so and this page narrows to your share of it.
A job title is a bundle of tasks bought together, and no two people hold the same bundle. Nothing is sent anywhere — it stays in this browser.
Read all 4 tasks in full — direction, reasoning and limits →
Recent changes#
United States. The statistics bureau projects employment of cartographers and photogrammetrists to grow 7 percent from 2025 to 2035, from 14,700 to 15,700, much faster than average, with about 900 openings a year. The page does not mention artificial intelligence. It counts jobs for one country and a small occupation; GIS technicians are counted in a different category.
A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.
United States. The statistics bureau projects employment of surveying and mapping technicians to grow 6 percent from 2025 to 2035, from 58,800 to 62,200, with about 6,900 openings a year. It says increased demand for map information will require technicians, and that although drones and other advancements make some aspects of survey work more efficient, technicians will continue to be needed to review and interpret the output of these technologies to ensure accuracy and completeness. It does not mention artificial intelligence by name.
A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.
Global. Microsoft's open dataset of building footprints extracted by machine learning from imagery covers 225 regions. Its README says its metrics show the data meets or exceeds the quality of hand-drawn footprints, reports precision above 92% and recall between 70.9% and 85.9% by region, and tells users never to import the data into OpenStreetMap without first checking local quality, which varies between rural and urban areas and terrain. The company describing its own dataset.
A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
A controlled experiment with 26 participants comparing fAIr, an AI-assisted building-mapping environment from the Humanitarian OpenStreetMap Team, with JOSM for manual mapping. Manual mapping was faster and more accurate overall, mainly because experienced contributors performed well; fAIr reduced differences between novice and experienced contributors but produced AI-related errors, most commonly merging multiple buildings into a single footprint. Volunteer mapping in two study areas; the findings informed a new version of the tool.
A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
A benchmark of 50 Python-based geoprocessing tasks derived from real GIS problems. Proprietary models such as ChatGPT-4o-mini achieved 95% workflow validity while smaller open models such as DeepSeek-R1-7B reached 48.5%; tasks requiring deeper spatial reasoning, such as spatial relationship detection or optimal site selection, remained the most challenging, and the authors call for rigorous evaluation before claims about full GIS automation. Valid workflows are not the same as correct results.
A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
France. The national mapping agency says its large-scale land-cover map (OCS GE), produced with AI from aerial and satellite images since a project launched in 2022, now takes one year per department instead of three — a production time divided by three — and that the technology halved the production cost of this geographic data. The agency describing its own production; it says nothing about staffing.
An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
A study that integrated a language-model agent into QGIS so users can run spatial analysis with natural-language commands, evaluated on over 100 tasks at three levels of complexity. It reports a high success rate in tool selection and code generation for basic and intermediate tasks, while challenges remain in achieving full autonomy for more complex tasks. Tested by its developers in a research setting.
A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
What this means for you#
If you are starting out, expect bulk digitising to be done by models, and build what they lack: spatial reasoning, checking data against the ground, and knowing when a machine-made layer is wrong.
Expect to spend more time correcting and validating machine output and scripting analyses with AI help. The judgement about what the data means and whether it is fit for use stays with you.
Your options#
Four directions, each with its real constraints and one thing you can test this week. Continuing as you are is a legitimate choice — it just has to be a chosen one.
Move from tracing features to checking machine-made layers
Mapping agencies are moving extraction to models, and even vendors tell users to check the output locally.
Correcting model errors, such as merged buildings, can take as long as tracing them.
Compare a machine-made building layer with imagery for one neighbourhood you know and count the errors.
Stay in spatial analysis and data quality
Benchmarks find spatial reasoning the hardest part for AI, and technicians are still needed to review output.
Analysis roles often expect programming skills.
Give an AI assistant a site-selection question you have solved before and check each step of its answer.
Move into training and evaluating GeoAI models
Model-based extraction needs labelled data, local tuning and accuracy checks that depend on mapping knowledge.
It requires machine-learning skills on top of GIS.
Try an open AI-assisted mapping tool on a small area and note which features it gets wrong.
Common questions#
It is taking over bulk feature extraction, not the whole job. A national mapping agency cut its land-cover production time to a third with AI, but experienced manual mappers beat an AI-assisted editor in a controlled test, spatial reasoning remains hard for AI agents, and machine-made data still needs checking. The US projects cartographer employment to grow 7 percent from 2025 to 2035.
We do not answer that with a number of years. Watch whether your organisation starts from machine-made layers rather than imagery, and how much of your time moves to checking them. Those tell you more than any date.
At scale, yes — Microsoft publishes machine-made footprints worldwide — but quality varies by place, and in a controlled test an AI-assisted editor most often merged several buildings into one outline.
It can write and run routine geoprocessing workflows, but benchmarks find tasks needing deeper spatial reasoning, such as site selection, the hardest, and researchers warn against claims of full automation.
What these judgements rest on#
1 of 4 task judgements on this page are backed by a verified event and 3 are platform inference, each labelled where it appears. Behind them sit 2 technology dimensions, a reconstructed trajectory since language models reached the public, and 7 verified events.
See which technologies, how it got here, and the method →
Where it sits in the official classification: skills, knowledge, related jobs →
Other roles in the same function#
A company divides its work into functions before it divides it into jobs. These sit in Core delivery (industry-specific) alongside this one — a fact about org charts, not a judgement that they are similar or that they are changing in the same direction.
Content moderator · Registered nurse · Radiologist · Pathologist · Radiographer / radiologic technologist · General practitioner / primary care doctor · Surgeon · Care worker / nursing assistant · Counsellor / therapist · Psychologist · Dietitian / nutritionist · Pharmacist · Pharmacy technician · School teacher · University lecturer · Chef / cook · Electrician · Plumber · HVAC technician · Architect · Interior designer · Civil / structural engineer · Electrical engineer · Chip design engineer · Mechanical engineer · Industrial engineer · Quantity surveyor · Journalist · Editor and proofreader · Translator / Interpreter · Interpreter · Retail cashier / shop assistant · Bank teller · Medical assistant / clinic assistant · Waiter / restaurant server · Construction worker · Auto mechanic / vehicle technician · Medical laboratory technician · Physiotherapist / rehabilitation therapist · Firefighter · Police officer · Farmer · Social worker · Dentist · Dental hygienist · Veterinarian · Librarian · Welder · Sonographer