Kitaab

The Chain of Causality Unravels

note

published 2024-01-07 15:19

updated 2024-01-07 16:53

When I was younger, I had the bewildering realization that pretty much everything in the built world was made by someone. It sounds so obvious as to not warrant further thought. How else would anything have happened? There is still wonder in the world for those who choose to see it, I 'spose. I do mean everything. From chairs to clothes to floor to food. Everything is manufactured, packaged, shipped, assembled, constructed; Engineered. Eventually, everything gets consumed. But it always starts somewhere.

This is really just looking at Causality as a hyper-object. It reminds me of the first time I heard the word Sonder. Instead of everyone's personal lives, you recurse down the processes that people have put in place and are a part of, that led to the creation of, basically everything I interact with. Even when I'm out climbing, the nature reserves are preserved and looked after by humans, the routes are developed and bolted by people. Even the trails are excavated from entropy by people. Not until I go far into the remote wilderness do things appear without much human intervention. Appearances are deceiving, for Climate Change impacts even the farthest reaches of our planet.

Humans were always at or near the center of all of these operations. A human plants the seed, or harvests the fruit, or designs the machines, or schematics. No matter what angle you look from, a human isn't too far from the action you choose to inspect. Even hugely automated processes require human oversight, quality assurance, or packaging. This causal chain with a human in the middle is finally beginning to unravel. With the rise of LLMs, we're seeing automated non-human agents that, for the first time, can push a human out of the loop entirely. As they grow in complexity, perhaps it will be several orders of magnitude of "process discovery" before we see a human involved.

Indeed, we've all heard about The Algorithm, and it's indecipherable desires. But all Big Algorithms are tuned by humans, their fitness function determined by people. Individual actions may not be observable, but by and large, why it does what it does is determined by the corporation that built it. As agents learn to spin up other agents, and give them their own script of actions to operate autonomously we may not be part of the loop at all. We become part of the world on which Causality operates.

I fear for the spam-scams generated by a genetically evolved LLM agent which uses it's crypto-wallet value as a fitness function. We're not quite there yet, but the pace of development in AI is staggering. It's already become weird to look at an image and decipher it's intent. At least for now, Who made it is still a valid question to ask. Even if it is a bot, someone directed the bot through some process. Soon though, who made it and why will be impossible to answer.