I Asked AI to Predict
My Weight a Year From Now

I asked Claude a simple question. If I eat 2,400 calories a day starting in August, where does that put me by next June. The answer came back more complicated than I expected.

I’m actually writing part of this from the Bahamas, four days into a trip that will not be doing my calorie count any favors. So the starting weight isn’t some idealized number. It’s 187, which already assumes I come home a few pounds heavier than when I left.

The first version was clean. Take my current deficit, hold it flat for a year, run the math. The line went straight down, no curves, no mercy. By June 2027 it had me at 141.3 pounds. I’m pretty sure my wife would find me repulsive at anything close to that number, and she is not a woman who keeps opinions like that to herself.

The problem was the model assumed my body stays exactly the same while it gets smaller. Same metabolism at 141 as at 187. Anyone who’s actually lost weight knows that’s not how it works. The lighter you get, the less fuel you burn just existing, and the deficit you started with quietly shrinks whether you notice or not.

So we rebuilt it. Same 2,400 calories a day. Same starting weight. But this time maintenance drops about ten calories for every pound I lose, recalculated weekly instead of frozen on day one.

Then I threw it a curveball. Assume I gain two pounds between November 1st and December 31st, because I already know what’s coming and pretending otherwise has never once worked for me.

This version looks like an actual body instead of a spreadsheet fantasy. Fast in the fall, a small bump for the holidays, then a long slow taper that lands around 162 by next June. Not 141. Not a straight line. A curve that gets harder to keep pushing the closer it gets to done, which tracks, because that’s exactly how this has always felt.

I don’t fully buy 162 either. Somewhere around March I already expect to hear my wife tell me I’m getting too thin, and when that happens I may just hold steady for a while instead of chasing the model down to the decimal. We’ll see how I actually feel when I get there.

I also picked 2,400 on purpose, and not because it’s the tightest number the math would allow. It’s about 250 calories higher than what I’ve actually been averaging since June. I know myself well enough to know that focus fades. Something always gives a little a few months in, and I’d rather plan for that now than pretend it won’t happen. The goal was never to nail a calorie target. It’s to keep tracking, every day, no matter what the number says.

But I like having a number written down somewhere I can’t quietly revise later. Next June, I’ll hold this post up next to whatever the scale actually says.

For now: the assumptions are 2,400 calories a day, chosen with room to spare, an estimated maintenance around 2,900 based on my own tracked burn, a ten calorie adjustment per pound lost, and one honest concession to Thanksgiving. We’ll see how close AI actually gets.