AI Mapping

I'm starting to explore a few AI related avenues for a project I'm going to be involved in for the next few months. 

I'm interested in the ethics of using AI when teaching geographical topics, and the implications for students' understanding of the world if they are exposed to images, text and other information which have been AI generated.

So far, we have had the launch of several large language models (LLMs) such as Chat GPT and a number of image models such as Dall-E and Midjourney. The map below was actually generated by Adobe Creative Cloud / Firefly for Education, which I have access to via school.

The prompt was: "A map of Europe, with the letters A and I hovering over it in gold with rainbow edges".

Adding a Structure Reference and regenerating created this output.

Quite a few people have shared how they use prompts to elicit certain outcomes from LLMs in a geography context. But should we be using it in our teaching and if so, what disclaimers should we display when we have used it? 

One ethical issue for me that should be considered is that of plagiarism, and the use of authors' works to 'train' the models without their permission. Using AI isn't developing students' geographical knowledge or imagination.

There are also issues with image generation. Models can't handle the fine detail of letters and other graphical content, or the fine details such as how many fingers on a person is too many.

There are some early map creators that have been developed and made free of charge. One is called PROXI, but doesn't seem to do a great deal that a simple tool like ArcGIS Online could do with a few clicks anyway.

A guide to Proxi is here.

Here's an output: a map of second hand book shops in Iceland - I made a map of Iceland to test it out.

powered by Proxi

There are going to be further developments happening at pace in this area one imagines, which will make it hard to keep up, but at the heart of it is the question of if and how we should be using this with young people, particularly when we are linking them with geographical information.

Mariana Mazzucato is professor of economics at UCL, and director of the Institute for Innovation and Public Purpose. 
She wrote this piece for the Guardian on the growing environmental costs of AI due to the energy needs of data centres.

This is another issue to consider - our online lives are having an impact which is not obvious. Even posting this blog post, and you taking time to read it right now has had an environmental impact and a carbon cost. Storing and caching this webpage for an inderterminate period of time will continue that impact.

AI's models use vast amounts of electricity and computational power.

Rather than come up with solutions to the climate crisis, it is likely to create further disinformation.

I also downloaded a keynote presentation from the recent EuroGeo Conference which was given by Nico van der Weghe from Gent University. It includes some useful ideas.

Please have a think and send through any relevant links.

Thanks to Bob Lang for sending a range of links already, which I will perhaps add in a follow-up post when I get a moment.

To finish, I like this sentiment from Joanna Maciejewska

"I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes".