Antonio Antoine also has something to say about generative AI in cartography. As I did in my piece about generative AI, he takes as his starting point the question of gatekeeping; unlike me, he’s coming at it as a working mapmaker. He takes the question at face value: who and where are the gatekeepers in cartography—for example, access to data, education, software licences—and asks what (if anything) generative AI can do about it. In the end he comes down on the side of accountability: “If one is not trained in geography/cartography, where does accountability step in to correct bad maps? Are formal cartographers supposed to police the internet? Like gatekeepers? Why not nip it in the bud at the source? […] As far as I am concerned, genAI in cartography is a race to the bottom. There will be little to no accountability in these maps as there is very little ownership in them.”
Tag: AI
Lee Schwartz on That Messed Up Map of Africa
Lee Schwartz, current CEO of the American Geographical Society and former Geographer at the U.S. State Department, has some words about the U.S. government map that messed up Africa at a conference last month (previously).
Maps are occasionally used that have an incorrect border or outdated name. A word misplaced or a typo can lead to a serious misunderstanding between both friendly and rival governments. Careless human mistakes, or a lack of guidance and oversight, are usually the reason. In this case, blame can apparently be placed firmly on the careless use of artificial intelligence rather than castigating an entire diplomatic corps with incompetence. The fact that this map made it to the light of day may be more reflective of failures and shortcomings in our institutions charged with representing U.S. foreign policy than a cartographer’s slip of the pen. […]
In international diplomacy, every recognition decision, every treaty map, every territorial dispute, and every official representation as rendered in maps depend on getting geography right. While software can presumably generate maps useful for a variety of purposes, geographic literacy is still required to provide historical understanding, socio-cultural context, and precision in conflict resolution negotiations. Not accurately locating or naming countries that can be potential allies for partnership agreements, investments, or security cooperation reflects a low priority placed on understanding them—and their unique people, places, and relationships.
This seems . . . diplomatic.
AI-Generated Wildfire Maps Misrepresent the Fires’ Extent
AFP fact-checks fake AI-generated maps of the Canadian wildfires that have been circulating on social media, many of which overstate the severity and spread of the fires, including hallucinating wildfires in places they don’t exist, for engagement. [Charles C. Mann]
‘Geospatial Trust Is Hard to Win and Terrifyingly Easy to Lose’
Ed Parsons, who used to work at Google, has some words about the Google Earth AI image generation debacle. “Veterans of the Geo organization understood an unwritten rule that took decades to build: Geospatial trust is hard to win and terrifyingly easy to lose. Google Earth is relied upon by human rights investigators, journalists, emergency responders, and courts of law as a baseline truth layer of our planet. Once you train the public to accept that the satellite view can be whimsically altered by typing a prompt about a ‘lakeside cabin’ or a ‘futuristic metropolis’, you destroy the integrity of the platform. If anything on the screen can be AI-generated, then nothing on the screen can be trusted.”
Previously: Google Earth Goes Nano Bananas (which has been updated a few times since first publication).
Google Earth Goes Nano Bananas
It’s been possible to use 3D models in Google Earth for decades. But Google’s announcement yesterday its AI image generation model, Nano Banana 2, can be used to create custom images and 3D rendering inside Google Earth is getting some pushback, and not necessarily just because it’s AI.
Henk van Ess is concerned that realistic AI-generated images in the context of Google’s satellite imagery layer will be weaponized in a way that undermines our ability to determine the truth.
Google spent twenty years building the reference the world checks against. Today it added a button that makes things up. […]
Google’s own description of the “fun feature” is that Nano Banana “creates concepts grounded in the real world.”
Grounded in the real world means the invented thing is welded to genuine coordinates, drawn on genuine imagery, often in the same colours and the same light and at the same angle as the picture beside it. […]
A government official wants a strike to look bigger than it was. A faction wants a hospital to look flattened, or intact, depending which one is holding it this week. A troll wants forty thousand reposts before lunch.
This morning all three needed a screenshot, a second browser tab and a little patience. Tonight they need a sentence.
Van Ess’s point is that while it has always been possible to falsify satellite imagery, it just got a lot easier. He also discovered that Google’s safeguards against creating images on harmful topics did not prevent him from doing so.
Update: 404 Media: “Google sent 404 Media a statement saying it was ‘rolling back this feature in Google Earth while we work on implementing stronger guardrails.’”
Update #2: Google’s blog post has been updated:
We know that people uniquely trust Google Earth for a reliable view of the world. We’ve seen geospatial professionals using this feature for a range of useful purposes, however we’ve also seen people sharing screenshots of generated imagery that appear to violate our policies. So we’re rolling back this feature in Google Earth while we work on implementing stronger guardrails. It’s important to note that generated images didn’t appear in the main Google Earth experience for others to see and were watermarked as AI generated.
Update #3: Engadget coverage.
Update #4, 3 Aug: See also BBC News coverage, in which BBC Verify tested Google’s AI checkers against the AI-generated Google Earth images and discovered that “it was possible to circumnavigate these checks and trick Gemini into saying these fake Google Earth images are real. Tests using external AI detection tools also in some cases failed to identify the Google Earth AI-content.”
Mislabelled, Misplaced and Misshaped
The Guardian: “A US government map of Africa mislabeled every country during a state department presentation at a global conference in Brazil this week, causing a stir among attenders who took screenshots and posted them online.” Not only were the countries mislabelled, they were misplaced and misshaped; as Wonkette puts it: “They’re all shaped wrong, and they’re also wrong in all the other ways of geographical wrongness.” At least the outline of Africa was okay. Not that AI is required to make a map so perfectly wrong, but it appears to have been used in this case.
Gatekeepers, Luddites, Haters and AI-Generated Maps
I’m opposed to generative AI and all its works. According to some leading names in the cartographic community, that makes me a gatekeeper, a Luddite, and an irrational AI hater.
So, generative AI has come to cartography. The ICA’s Commission on Map Design has been posting examples of maps produced using generative AI tools, and there’s a website highlighting the ways you can use ChatGPT to generate maps using natural language prompts. The proponents of this stuff have been making some fairly inflammatory statements in its defence, such that I can only imagine they’ve been getting some grief about using it and are pushing back against it. Hence the stuff about gatekeepers, Luddites and haters. These are epithets I’ve seen thrown around before by pro-AI advocates in other fields where the use of generative AI has proven even more controversial. They’re not new. They’re also not worth taking seriously.
Continue reading “Gatekeepers, Luddites, Haters and AI-Generated Maps”Google Maps AI Updates: Ask Maps, Immersive Navigation
Google just announced a couple of fairly major Gemini AI-powered updates to Google Maps. Ask Maps is a a chatbot that produces personalized responses to questions—essentially an intermediary that sifts the data so you don’t have to, taking into consideration your known preferences (with all that entails: not necessarily good). Immersive Navigation is a 3D mode full of suggestions:
When it’s helpful, Maps will highlight critical road details like lanes, crosswalks, traffic lights, and stop signs to help you make that turn or merge confidently. This spatial understanding of your route is made possible with help from Gemini models, which analyze fresh, real world imagery from Street View and aerial photos to give you an accurate view of things along your route, like landmarks and medians.
Includes voice guidance in more natural language and explaining the pros and cons of alternate routes. All of which requires that the underlying map data be accurate and up to date. We’ve already seen what happens when people blindly follow GPS/satnav driving directions that are in error or out of date; if anything people have proven to be more even credulous with AI chatbots. So we’ll see how this goes.
AI Crawlers and the Cost of Geospatial Infrastructure
Bill Dollins reacts to Gary Gale’s experience with AI crawlers taking down his mapping project (previously), and what that portends for the open geospatial web. “On its own, this is a small incident. No critical infrastructure failed. No global service collapsed. It is, however, a revealing stress case. It shows how open geospatial infrastructure behaves when exposed to a new class of demand. That demand is continuous, automated, and indifferent to the social and economic assumptions that shaped the system in the first place. This is not an isolated story. It is an early signal of a broader shift already underway.”
‘And Then the Bots Came’
(AI) Bots Ate My Map Tiles: In which Gary Gale discovers that his Vaguely Rude Places Map’s 200K-map-tiles-per-month plan is no match for the hammering delivered by AI crawlers.
Londonist Asks ChatGPT to Draw Maps
“The shortcomings and possibilities of generative AI are, of course, well chronicled across a million op-eds. I could write at length about the dangers or opportunities the technology presents,” writes Matt at Londonist. “But this is a newsletter about London, and I’m still in a silly holiday-season mindset. So all I’m going to do today is ask AI to draw some historical maps of the capital, and then take the p*ss out of them. Popcorn at the ready . . . ” It goes about as well as you’d expect: “terribly,” with results “as crazy as a yacht of numbats,” with labels “so bizarre that I don’t know where to begin.”
Review: GeoAI
Save some room on the AI bandwagon for ArcGIS. This seems to be the central message of GeoAI: Artificial Intelligence in GIS, a slim (only 120-page) volume of articles and posts that previously appeared, for the most part, in Esri blogs and publications. They highlight examples and “real-life stories” of how Esri’s machine- and deep-learning tools have been successfully applied in the public, private and non-profit sectors. At a moment when “AI” is invariably a synecdoche for the awfulness that is generative AI, which I will not litigate here, it can be a challenge to remember that machine and deep learning tools, which have been included in ArcGIS since 2008, have all kinds of applications and benefits. (See Esri’s pretrained deep learning models for examples like feature detection, land-cover classification, and object tracking; see also their GeoAI landing page.) Calling these tools “GeoAI” strikes me as a way to package them to appeal to decision makers who are speedrunning their AI rollout, for better or worse. It’s those decision makers that this book is targeted to. Esri has something to sell them: this is the pitch.
I received an electronic review copy from the publisher.
GeoAI: Artificial Intelligence in GIS
ed. by Ismael Chivite, Nicholas Giner and Matt Artz
Esri, 2 Sep 2025, $40
Amazon (Canada, UK), Bookshop
AllTrails and AI-Generated Hiking Trails
Last month the hiking app AllTrails announced AI-generated (“leveraged”) custom routes as part of a new premium membership plan, and some people are worried about it. According to the National Observer, AllTrails and other hiking apps have gotten hikers into trouble because they rely on crowdsourced trail information, which isn’t necessarily official or safe. Given generative AI’s track record for producing spectacularly erroneous results, there would appear to be some cause for concern. Except that “AI” has become a marketing buzzword that covers a lot of computer stuff, from less problematic machine learning (which is what I’d expect in this case) along with more problematic generative AI/large-language models, and AllTrails isn’t indicating which flavour they’re referring to (because: buzzword). And as the National Observer points out, “These problems already existed before the AI was added.” To be sure, generative AI is a blight on human civilization, but let’s be clear about our targets in this case.
AI Chatbots and Geolocation
AI chatbots don’t have the best track record when it comes to accuracy. They appear to struggle with geolocation too, as Bellingcat discovered two years ago in a test of OpenAI and Google chatbots. Bellingcat has now tested them again, this time putting 20 large-language models to work on 25 travel photos to see if things have improved, with Google Lens reverse image search as a control. The result? A few ChatGPT models outperformed Google Lens, but not by much; the rest were worse. Details at the link.
(Update: Bellingcat’s coverage goes in quite a different direction than reports last April highlighting ChatGPT’s “scary” ability to pinpoint locations from photographs, largely because it compares it with existing non-AI reverse image search. Privacy risks may not depend on the kind of technology analyzing the photo, in other words.)
Some Google Maps Updates
Google Maps imagery updates include improved satellite imagery thanks to an AI model that removes clouds, shadows and haze, plus “one of the biggest updates to Street View yet, with new imagery in almost 80 countries—some of which will have Street View imagery for the very first time.” The web version of Google Earth will be updated with access to more historical imagery and better project and file organization, plus a new abstract basemap layer. [PetaPixel]
Meanwhile, The Verge reports that Google Maps is cracking down on business pages that violate its policy against fake ratings and reviews.