NATO's drone-predicting AI, called No. 13, gives troops a map of tomorrow's strikes - Business Insider

NATO has built a machine that tells you where the drones will be tomorrow, and it is called Model 13. Not Aegis, not Sentinel, not some Capitalised Acronym with an X in it — thirteen, which is the number you arrive at when the naming meeting runs past four o'clock and somebody reads out a spreadsheet tab. It comes courtesy of NATO's Joint Analysis, Training, and Education Center, an institution with 'Education' in the title, now teaching a model to guess which drone visits which substation. A 'map of tomorrow's strikes', Business Insider calls it — comfortably the most beautifully ominous phrase anybody has put above a defence-technology story this month.

Consider the input list, because the input list is where things get genuinely endearing. Previous Russian drone activity. Drone types. Deployment timing. Weather. Terrain. Distance to the line of contact. Six variables, meticulously modelled, and not one of them is 'a colonel in Rostov wakes up spiteful and empties a warehouse because procurement needs to justify the quarter'. You can pin the wind over Zaporizhzhia at four in the morning to three decimal places and still be comprehensively wrong, because war also runs on caprice, boredom and paperwork, and none of the three appears in the feature set.

And the output is tomorrow. One day. Which is, coincidentally, roughly the amount of warning a commander previously derived from looking out of the window and listening. Presumably Phase Two stretches it to Wednesday. Meanwhile 'predictive AI for the frontline' is now loose in the world and will be copy-pasted into defence acquisition documents from here to 2030, with the word 'expected' quietly deleted somewhere around page forty. Somewhere a staff officer is already drafting a slide titled Decision Advantage. The drones, tragically, do not read slides.