*tl;dr: Trying to get my head around the implications of the arrival of AGI on the future of the human race.*
I was watching “[Singularity or Bust”](https://www.youtube.com/watch?v=owppju3jwPE&list=WL&index=8&t=0s) the other night.
It’s about the (nearly) inevitable advent of Artificial General Intelligence.
Warning: it’s a very intense documentary. It’s about 40 minutes long.
<iframe width="734" height="315" src="https://www.youtube.com/embed/owppju3jwPE" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
One of the big questions that the film raises is:
> **What will happen to humans when we create machines that can truly think for themselves?**
The answer, it seems, is that there is a new addition to this image.
It’s on the far right side and it’s the being that is non-human that we create.
**Today, AI is Narrow**
Today’s artificial intelligence programs- those that do everything from helping you get from point A to point B to telling the Starbucks barista which mobile order to prioritize- are wondrous, but they are narrow purpose tools.
They are good for that one thing, but not good for everything.
Humans, however, are really, really good at many things.
The reason we are so good at so many things is, in part, because our brains are pattern recognition machines. They are able to take vast quantities of disparate ideas, concepts, and facts and categorize them to make sense of the world.
In fact, there are now university-level research programs that affix cameras to the heads of babies and toddlers in order to figure out how humans are able to learn so much about the world so quickly. We are [“The Ultimate Learning Machines.”](https://www.wsj.com/articles/the-ultimate-learning-machines-11570806023)
If you want to see world-class pattern recognition in action, [here’s a time-lapsed video of Rohit Bhargava](https://youtu.be/5jkn9hsnqdk) using his “haystack” method for recognizing emerging societal trends.
<iframe width="734" height="315" src="https://www.youtube.com/embed/5jkn9hsnqdk" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
[Rohit](https://www.rohitbhargava.com/) is the author of the world-renowned [Non-Obvious Guides](http://www.nonobviousbook.com/) and the founder of [IdeaPress publishing](http://www.ideapresspublishing.com/), that provides an entire line of easy-to-understand and practical guides for smart people.
(He also has one of the [best newsletters you’ll read.](https://www.rohitbhargava.com/))
**Tomorrow, AI is General**
All of that is well and good.
Right now, Rohit is better at looking for trends across broad swaths of data than an AI algorithm.
However, tomorrow, just like a baby who doesn’t know a dog from a cat but grows up to be a nuclear physicist, the machines we build, powered by the software we create, are going to be able to see those general, abstract trends.
Then, they will be able to see them faster and faster.
They will be able to learn faster and faster, applying that knowledge in new ways to solve different, bigger problems.
On the one hand, this could be great.
On the other hand, it could be terrible.
The thing is, we don’t know what the outcome of AGI will be, but we do know that we will have created beings that will be superior to us.
Homo Erectus created Homo Sapiens.
The latter outlasted the former because of its superior ability to adapt to a changing environment.
Homo Sapiens is creating AGI.
Then, we’re not sure what happens.