US pet industry spending

It is a curious fact, and one that most people find nothing short of bewildering once they think about it for more than thirty seconds, that the amount of money Americans spend on pet food, toys, and veterinary services rises and falls in almost perfect sync with the number of papers someone publishes on arXiv about artificial intelligence, which is to say that as we collectively decided to lavish more affection on our dogs and cats, we also collectively decided to teach machines to think. One would have difficulty inventing a less plausible pair of dance partners, and yet here they are, moving in lockstep across fourteen years as though choreographed by something with a genuine sense of humour about irony.

The truth, which is somehow both more comforting and more unsettling, is probably that both pet spending and AI research papers are symptoms of the same underlying fever: disposable income and technological enthusiasm among the same educated, affluent slice of the population, swelling and contracting together with economic confidence and employment in knowledge-sector jobs. When the economy felt robust between 2015 and 2019, Americans spent roughly 131 billion dollars annually on their pets—an amount roughly equivalent to the entire GDP of Iceland—while simultaneously flooding arXiv with papers on neural networks and transformers, because apparently prosperity makes us want to both pamper our companions and build artificial ones. The third factor, of course, is probably just that growing numbers of well-funded people in wealthy countries tend to do more of everything at once, including spoiling their actual pets and theoretical ones.

What emerges is not that pet spending causes AI papers or vice versa, but rather that both are emanations from the same economic and cultural moment, rising together like two equally baffled passengers in the same elevator, neither causing the other to rise. We are pattern-seeking creatures who live in a rich enough time to pursue two seemingly unrelated obsessions simultaneously, which is either profound or merely human. The data, in its quiet way, suggests the latter.