Growing A Bigger Ox


This one is going to be all over the place, don’t say I didn’t warn you. We start with the story of my late grandfather. He was born in a village in the state of Gujarat, where for much of his life, it did not have widespread indoor plumbing and electricity. His brothers worked in the farms so he could get a better education. After being the first in his family to finish high school, he went on to get a Bachelor’s in physics. Eventually he found himself in Louisville, Kentucky where he studied mechanical engineering, then at Bucknell where he finished his Master’s. Afterwards, he worked in Tuckerton, New Jersey. Only to make full circle back to India, something about serving in the freedom movement as a high schooler and wanting to bring his new knowledge back home. Not only that, dude took a boat back via Europe and Africa. He somehow managed to be a figurative FOB in the States, and a literal FOB back in India.

A lot has changed with that village. Nowadays, there’s a school there with a bunch of computers. All of our relatives’ homes have indoor plumbing and ample power. The roads to get there are paved. But even that is fairly recent, some of it within my lifetime. I still have some memories of the time he and my grandmother took me there as a young child. We drove his car that somehow always had cockroaches on bumpy dirt roads. The homes of our relatives had earthen floors. And there were two sheds in the back, one where you got one bucket of hot water to take a bath with and its neighbor with a hole in the ground. I’m not trying to paint the picture of some poor rural village. We had running water and could fill ample buckets of cold water; the house had lights. But it was definitely a significant step down in comfort from what I was used to in the Northeast.

I’ve gone back to that village several times throughout my life. And every time I’d get reminded how driving around here isn’t an old phenomenon. For generations a bullock cart was the common mode of transport. A bull would help plow the fields, and a bull would help pull the cart to where they needed to go.

Now to let my goldfish brain take over and pivot. Recently I was listening to a talk (part one and part two) given by the legendary American computer scientist, Grace Hopper. Towards the end of part one she gives an example of how before the advent of modern mechanized industrial equipment, we would use oxen and horses to do work. If we found something too large or heavy to move, we wouldn’t go back to the farm and try to grow a bigger animal. We’d instead get another ox, and hopefully the combined effort of two would output enough force to move the big thing.

It got me thinking. Cue pivot. We have more compute power than we ever have before in human history. We literally put a man on the moon with a computer with less compute power than the calculator I had in middle school to learn algebra.

Again, we have more compute than we have ever had in human history. And yet our software has never felt buggier. I have devices running macOS and Linux (Ubuntu, Mint, Manjaro, and Arch [I use it btw]). I have a device running W@#$%^s. Between personal use and work, I have a device running iOS and Android. Sure I’ve cast a wide net, but it’s not a good sign that on a weekly basis I consistently run into something crashing or freezing across every category of device.

Now I am by no means an expert in artificial intelligence. I started my career big-data-adjacent, I’ve done some company hackathon projects with TensorFlow, and I was briefly pursuing a Master’s in machine learning. Not much, but enough to understand that the end result we see as users is really just an ungodly amount of data being fed to code that implements math and statistics.

Five hundred billion dollars is a lot of money. Even in the great state of Texas, my power bill is already pretty high. Despite the dot-com bubble popping, the internet continued to grow and became integrated into our daily lives. On the other end of the spectrum we have blockchain. For better or worse, like both the web and cryptocurrency, I don’t see generative AI going away. But half a trillion dollars? I use Copilot at work, ChatGPT in my side business, and Anthropic for personal projects. It’s effectively an intern that is wrong most of the time, but every once in a while does a good job. It adds value, but not five with eleven zeros value.

Adjusted for inflation, that is almost double the cost of the moon landing.

Again, not an expert. Thanks to the advent of cloud computing, we have a lot of oxen. I really need someone to explain to me how spending eleven figures to make an even bigger ox is going to make the virtual intern better.

Idk, maybe it’s time I start playing around with Ollama.

Edit: Jan 27 2025, I guess DeepSeek kinda proved the point, similar performance to OpenAI’s reasoning model, but supposedly developed at a fraction of the cost and compute.