I've started using ChatGPT, DALL-E and CustomGPT bots, NotebookLM and other AI platforms, to explore the use of AI in wargaming, role-playing, peer reviewing and other areas. This page is a link to my posts on the topic, and also a holding point for documents that those papers reference.

The main article on using ChatGPT to simulate the whole of a matrix game is on Substack. The files referenced by that article are:

• Simple Russo-Ukraine Matrix Game Scenario
• First Playthrough of the Russo-Ukraine Game
• Russo-Ukraine War - direct narrative generation

Given the situation in the Middle East (Oct 24), I’ve also run an MGS on the Israel-Iran regional conflict.

NotebookLM from Google looks like it might be a better bet for a Virtual Urban Conflict Researcher based on my writings, as all its knowledge comes from what you upload, rather than any larger LLM. It also lets you upload upto 50 source documents, far more than the half dozen or so, but way below the 1424 items currently in my Zotero research database - so I think it's (at the moment) more of a presentation/reminder tool than a truly research one (although merging Zotero and NotebookLM would be one hell of a system.

The scene-stealer from NotebookLM at the moment is its ability to make a “podcast” from your sources. This is a one-click action, there is no ability to set any parameters and all the resulting podcasts are hence in the same cheery American style - but not bad for a first attempt. Here are the ones I've created so far:

• History of Urban Warfare (9min) - extracted by NotebookLM from the First Bites of Chapter 1 of the thesis - although it seems to really only focus on history, not historiography, urbanism or doctrine.

• The Urban Calculus (12 min) - a fairly typical (?) podcast review of a new game in style, extracted just from the TUC rules, designers notes, QRS and board. Bit miffed by the comments about TLAs and apparent complexity at the start!