Look at what work we do through computers.

Our loop is pretty simple:

1. Find something in the real world that we need to simulate fast or at scale
2. Find ways to create rough digital equivalents of it using manual data entry, automated data collection through sensors etc
3. Make computers run programs over the digital equivalents and figure out the results in the digital world
4. Assume that calculation in the rough digital world will be close enough to calculations in the real world

We make a basic digital equivalent of the world in the computers' memories. Usually by dumping data in files through a mix of manual entry and automation.

(into databases or a dump of structured files)

Then, we write a program to tell the computer what simulation to run on the data. That's all we've been doing with them.

Insurance company?

1. Create a rough digital equivalent of the real world by feeding customer's specific (and limited) parameters like age, location, medical history, test reports
2. Feed these billion records of 10 million customers into the digital representation (into a database)
3. Give the computer a model, and tell it to run a basic simulation to calculate the risk profile of every individual customer.

Chemical company?

1. Feed the live representation of the real world into the digital world through cameras (of good enough resolution)
2. Give the computer a model and tell it to keep watching all the feeds, and warn when it detects a fire

Weather prediction system?

1. Feed the live representation of the real world clouds taken through satellite cameras, into the digital equivalent
2. Tell a system to keep going through the latest feeds and predict winds, storms etc

You'll notice that we don't create a perfect model of the world. We only take in the parameters we think are most important. Because if we track each and every parameter to achieve 100% accuracy in the digital world - we will end up simulating the universe itself. We don't have that much compute, so we sick to rough good enough digital models, and spend out limited compute on that rough digital model.

And we do this through our current computers for 2 reasons:

1. Computers are fast. You can program lightning fast calculations. Eg: Your car's ecu cluster is continuously controlling the mix of fuel and air to keep the car running. Continuously monitoring the impact sensors and deciding if should trigger the airbags.
2. Computers can scale. Eg: Amazon scanning through the trillions of purchase records, figuring out purchase patterns.

We humans too do computing. In specific cases, we can do it fast (instincts are like FPGAs anyways). But we consistently fall short when it comes to computing at scale. The only way to scale for us is to use pen and paper to overcome out limited contextual memory.

As far as our brains go, biology has been evolving incredibly slowly. Then we found another way to scale beyond the cranium - computers.

And suddenly, problems have become solvable.