I did a quick calculation and got to around 500 queries per quarter pounder. Lot of guesstimation and rounding though, but I’m pretty sure I got close enough to know that you’re off by quite a lot.

Edit to add: I used 21.9kg CO2 per 1kg of beef and 4.32 grams per ChatGPT query for my rough estimate.

However that 4.32 number is already over a year old. Chances are it’s way outdated but everyone still keeps on quoting it. It definitely does not take into account that ChatGPT often “thinks” now, because chain of thought is likely as expensive as multiple queries by itself. Additionally the models are more advanced than a year ago, but also more costly and that CO2 amount everyone keeps quoting doesn’t even mention which model they used. If anyone can find the original source of this number I’d be very curious.

Do you have a citation handy for that?