What does the world look like by the early 2030s if the present pace of AI acceleration takes place?
Here is something that people are overlooking. Some of the technical people are aware of the pace things are changing. However, the average person is still asleep when it comes to what is taking place.
It is something that makes sense. Few frame it in a manner that penetrates the psyche of most individuals.
Perhaps there is a something that we can do to alter the narrative around this topic.
In this article I will go through some of the numbers and how it will affect things.
Often context is the most important criteria for comprehension. The data I will be using was compiled by David Shapiro in the following video.
Here are the number he came up with regarding the output of a single AI task and the timeline:
2026 - 3 hours
2028 - 8 days
2030 - 41 months
2031 - 58 years
2032 - 1,212 years
This is why we can look at the job situation, especially as it pertains to knowledge work as primed for disruption.
We know 2028 is now 3 years away. If we have an AI task equal to 8 days of human output, how many jobs can that replace? It will not take long for companies to start replacing people in massive numbers.
Another way to look at it is that a single AI task will have the same human output equal to almost 2 weeks worth of work.
Of course, this brings up a host of other problems outside just the job situation. The energy problem is only growing, with the thirst for electricity set to skyrocket. All involved in this at scale are hard at work on the problem.
The most recent solution appears to be moving a portion of the automated activities to space. This is something that could alter the entire equation.
Could we see AI becoming space based?
This is something that Elon Musk is getting more involved in discussing. That said, he is far from the only one going in this direction. Many are contemplating the potential.
If we consider the tasks required, why would something that is automated need to be on Earth. This is not only true for data centers but other manufacturing that doesn't deliver products back to Earth.
For example, the construction of the data centers along with all the components could be done from space (or the Moon). This means that launch would be less expensive and not have to deal with the gravitational force of the planet.
One challenge does become latency. This will negate some of the tasks that are moved to space. While this might seem to be a problem is just another layer to what we already have.
Autonomous vehicles cannot compute in the cloud. The inference must occur locally, on the vehicle. A slight delay could cause accidents, something this technology is designed to alleviate.
There will be operations that must go on Earth based data centers. Where we see a change entering is the fact that most operations can occur with a second delay. Does a prompt that provides feedback based upon reasoning require split second timing? If it is delayed a few seconds, that is likely acceptable.
The acceleration is going to change everything. Over the next 3 years, the world is going to completely change if the AI acceleration rate remains constant. The numbers simply bear this out.
