The Dow Jones is extremely weird in our modern world. The Dow Jones Industrial Average is supposed to track the performance of businesses over time. You would expect that if a $1T company loses 10% of its valuation that should affect the performance of the Dow Jones far more than if a $1B company lost 10% of its value, right? Instead the Dow Jones is price weighted, meaning a tiny company with a high stock price affects the index far more than a giant company with a lower stock price.

How does this make sense? Why would anyone create an index that’s supposed to track the performance of the economy but not take into account the fact that larger companies with low stock prices affect the economy more than smaller ones with high stock prices?

It begins to make much more sense once you remember the Dow Jones was created in 1896. A price weighted index is far faster to calculate by hand than a size weighted index.

Cutting corners like this reveals a previously common phenomenon that feels completely alien to someone born in the modern information age: compute starvation. I’m so used to the idea that complex calculations with large numbers can be trivially executed in nanoseconds that a world where that’s difficult feels foreign and exotic.

I think this is a great analogy to how I hope my children will feel in the decades to come, assuming we’re able to create aligned general intelligence. There are intelligence constraints all around that are invisible to us simply because the raw resource of smart person time is limited and can’t be funneled in vast quantities towards every possible problem. How this limits us is an interesting thought to toy with, because intelligence constraints have been baked into humanity since the dawn of creation. Intelligence constraints are so fundamental that seeing them feels like a “what is water?” situation. Just like the Dow Jones seems odd today in a world of abundant compute power, I hope much of what we create now will seem odd in a future world of abundant intelligence.

It seems like (with very broad brushstrokes) the way that we create almost all objects and software is limited by intelligence constraints. We try to design the fewest number of things that the most people can use. When we’re designing–to pick something random sitting on the desk in front of me–a pair of sunglasses, we try to design a single pair that will fit everyone’s head. Humans like feeling somewhat distinct so there are often lots of variations for colors and aesthetic styles, but the sunglasses are not generally custom fitted to your head because that requires time to tweak and design.

The same goes for software. There’s a limited amount of people in the world who can write code and using repeatable building blocks like libraries vastly speeds up projects. We try to create apps that the largest possible number of people can use with the least possible work. What if instead every app was custom generated exactly to your preferences and intuitions for exactly what you wanted to do?

Take something as simple as the phone app on an iPhone. To a young person who has grown up with “Contacts” and doesn’t have anyone’s phone number memorized the most intuitive starting page is the “Recents” or “Contacts” page. For an elderly person on the other hand this is often confusing. The most intuitive interface for them is usually the keypad page where they can directly punch in the number they want to call. Something as basic as this could be an on the fly decision made by a general intelligence where all the software you use is tweaked or created whole cloth to match your wants and needs without forcing you to go spelunking in technical documentation to discover the myriad flags and options you could set.

If the AI revolution goes well and doesn’t either peter out or kill everyone, I hope that we’ll be able to finally live without intelligence constraints. Where anything and everything can have vast quantities of intelligence funneled into it. Every protein and pathway in the human body can be explored. Every piece of information we need to learn can be precisely tailored and tied back to our level of understanding and current knowledge. We will no longer be starved for intelligence.