Jenga tower of abstractions

Abstractions are “things” the world denotes as truth. When something complex is abstracted, it is simplified, taken as fact, and then built upon.

For example, do you know where the world’s food comes from? We assume that the agricultural system works and trust that food is safe to consume. For a lot of people, food (specifically farming) has been abstracted away.

Another abstraction is code. Coding languages have a compiler that converts code into assembly. But in school, we don’t learn how the compiler works, we learn Java or Python and assume the compiler is correct. Assembly language has been abstracted away.

Abstractions are generally great. They free up time for us to innovate.

But is it possible that these same abstractions also create negative externalities? In economics, we learned that market pricing fails to factor in long term harms that aren’t visible in the present.

For example, will the cheap prices for food lead to cancers or heart disease? Will the decreased cost of LLMs mean increased hacking? Will social media lead to decreased long term productivity? These are all additional costs that may eventually have to be paid, whether it’s the cost of healthcare, the cost of dollars on cybersecurity, or the cost of a mentally slower workforce.

Enter the elephant abstraction, AI.

AI is looking to abstract work away into English (or your language of choice). The story of removing menial labor sounds great, but are there corresponding hidden costs?

If you google leaky abstractions, there’s an article explaining that when the compiler was built, it initially had a ton of mistakes. People would write code, and it would literally not work. Consequently, software engineers would have to go patch the compiler. The markets fixed these bugs quickly, because they had outstanding negative impacts in the present - e.g., the code wouldn’t run.

But when things look like they are working in the present, I can see how we might start building a jenga tower of abstractions.

AI is speeding up innovation. In Jenga terms, we are about to stack a lot more blocks at the top.

But like with all Jenga games, eventually, a lower portion of the tower starts wobbling from a slightly misplaced block. These are the negative externalities.

In life, we posit that technology will fix the misplaced block and keep the game running. If the price of carbon isn’t factored into electricity generation, we say that carbon capture technology will eventually be created. Yeah, maybe our food has a lot of pesticides, but even if it does cause cancer, technology will eventually cure that too.

And so far, it’s worked. If there is a negative externality, we and the market tend to notice things and fix the wobbly blocks before too many turns pass.

But today seems different.

What if the turns go by quicker, the tower gets built spectacularly tall, and the game ultimately finishes as intended. Scattered bricks all across the ground, and we have to restart the game.