While some online folks are arguing amongst themselves about the value of our little AI mapping experiment, continuing to label us as ‘slopmerchants’, AI proponents, and determining that we are incapable of understanding and accepting critique I thought I’d reiterate a few comments we’ve made elsewhere but which seem to have gone unheard.

If you begin your so-called critique threads with a statement such as “I don’t like AI maps” then proceed to simply say you don’t like what we’re doing you’re not actually offering anything useful. If you add personal and derogatory assessments of us, and our motives to that, you’re not helping move the discussion forward at all. In reality you’re starting an argument, then framing us, and our work in pejorative terms while continuing to try and provoke a response. As we’ve said many times, we’re simply not going to engage. We’ll be reporting our findings, what works, what hasn’t worked very well, where AI (and us, as operators) have failed, what works for cartography, what perhaps doesn’t (yet?) in good time. That’s the aim. That’s the point. That’s always been the point. And in that mix we’re now also likely to comment on the way in which people’s reactions to this project have verged on the intolerable at times. It’s been an emerging issue which is going to be important to reflect upon.

Today’s map is made using Generative AI because it’s a pictorial, or illustrative, map. I am fond of illustrative maps so I naturally gravitate towards this style. In looking around my home right now as I write this I can see Jasper John’s ‘Map’, Jo Mora’s ‘Yosemite’, Hal Shelton’s ‘Colorado – Ski Country’, The Tate Gallery ‘Tube by Tate’, James Niehues’ Ski trail maps, Stephen Walter’s ‘The Island’, Ruth Taylor’s ‘California’, MacDonald Gill’s ‘Wonderground’ map, Anton Thomas’ ‘Portrait of a Continent’ and ‘Wild World’ and many others. This style of cartography is not to everyone’s taste of course. But as a cartographer, when I see a theme entitled ‘Water Park Day’ I am naturally inclined to imagine a landscape where the map is a fun landscape surrounded by water in which characters are having fun. This is not much different to how Jo Mora envisaged the landscapes he drew. The difference is that I’m experimenting with Generative AI to assist me with the production of an artistic map. I can draw (though certainly not as well as Mora), and I’ve shared some of my hand-drawn maps before. But let’s say I don’t have weeks to put the map together, and maybe one can imagine a client demanding a map in very short order. In that sense, and for this purpose, Generative AI provides an opportunity. I don’t have copies of Tolkien’s maps but one could imagine him experimenting with AI to build his landscapes, and I’m sure many contemporary map-makers of fictional places are also doing so.

One criticism of this has been that this sort of art is derivative. Well, yes, to an extent, but so is most art. As humans we see things, and we build on the inspiration. This is really all an LLM is doing. I know people who openly admit nothing they do in their map-making is original and that they copy everything from somewhere. Yet they produce good work that is attractive, and widely praised. If creating work so derivatively is such a heinous crime then such people should be out of a job rather than being praised right? Instead, people turn a blind eye to some people’s practices, but if it involves AI, then that’s roundly criticised. Isn’t that double-standards? Other people in the industry are also using AI but they’re not admitting it, or explaining what parts of their work rely on AI. Maybe that’s where Ian and I have gone wrong in being totally up front about this particular project. We’ll never know. Though it is fascinating to see how some are evolving their work and using AI under the radar without any criticism.

Another criticism is that this will undoubtedly kill the careers of illustrators, and of cartographers, perhaps. Hasn’t this been an accusation of every technological shift in history? Again, some folks may find their ability to attract commissions is reduced. Or maybe they can harness AI in ways that they find acceptable and useful to assist in their design and production? Perhaps this is by training AI to understand their unique style and have the AI return work that mimics that style. This already occurs for the written word, but maybe Generative AI illustrations will develop in this way too. Ian has developed a lot of styles over the past few weeks and I intend to play with these a little more.

So, today’s map is a playful, illustrative map that imagines the world map as a water park. Generative AI is pretty good at these sort of highly saturated imaginary depictions. There’s plenty of maps from across the last few decades that take a similar approach. In particular, maps designed for children with bright colours, and lots of playful detail. They’re not trying to be serious. Just colourful fun. Nothing more. Nothing less. It’s designed to bring joy. Sometimes that’s all a map needs to do, and I like those sort of maps because I’ve made some of these by hand before too (remember Trump’s Ties?). I could have iterated the map more. I’m getting a little tired of these maps all having curved titles centred at the top. They sometimes look very similar but hey, if that’s a ChatGPT style, then that’s something recognisable right?

I hope you like today’s map too, but if not, that’s OK. It isn’t a reason to have a meltdown. Just move on, and save your judgment for when we bring our little experiment to a close and offer our reflections. And in the meantime remember the park rules – Be Kind, Stay Safe, Drink Water, and have Fun!

Original prompt (ChatGPT)