We still have some debate about the eventual impact of AI.
On one side are those who believe that AI is simply going to usurp anything. This tends to come from those involved in the industry, either using the tools to a large degree or creating them. Then we have those who believe a near status-quo will be achieve. Under this scenario, we will see improvement over time but nowhere to the level people are forecasting.
It is where the AI bubble and the believe we are in a hype cycle is coming from.
Personally, I believe this stems from the fact that people are not properly framing what we are dealing with. The comparison to the Internet doesnt really hold. That was networking, something that we know is slow to develop as evidenced by blockchain.
AI is a new form of compute. That puts it on a different level. It is resident in the digital world, at least to start. It is something that is going to follow the trends of the 1960s-1990s. There is a key difference.
In this article I will dive into some of what I foresee as emerging.
We see a lot of comparisons to the Industrial Revolution.
People are making the case that we are going to see a similar transformation of society. On this, I agree.
One thing that gets overlooked is the fact that the shift during that period was not absent issues. We saw many problems arising including labor conflicts that turned deadly. It would be great to say it happened only a few times but history shows this is not the case.
Society, or at least segments of it, had a tough time adjusting.
Here is the major differentiator: that occurred in spite of the transition occurring over a century. There were many decades to prepare. Generations were spurned during this time period. And yet, there were massive issued that were difficult to resolve.
What happens when there is a contraction?
This time, we are not looking at a century. Instead, it is only a decade or two. During this period, society will change completely.
How do things look? That is the multi-trillion dollar question.
The industry can get caught up in terms or metrics.
One of the things we see hotly debated is whether we have achieved AGI or not. Meeting some definition is not really important. What is vital is the capabilities of these systems.
A key component is the recursive training capability. This is what makes the acceleration a central part of any thesis.
Recursive training is the ability for AI to keep learning on its own, with no input from humans. It equates to 24/7 progress at a rate that only increases over time. Humans are limited by time. We also see these systems do things in seconds which can take humans hours.
The net result is a massive increase in output. This can be applied to whatever you desire. For the sake of our conversation, this basically breaks down to code.
Of course, the expansion continues.
Tesla, according to many, has a lead in autonomous driving that cannot be surmounted. The pace of data collection far exceeds that of its competitors. While this conclusion is certainly subject to debate, it does align with one of the key principles we are seeing in the AI world.
The more data that is amassed and processed, the better the system produced.
In other words, the goal is to get as much data as possible in a short period of time. The companies that are doing that tend to be gaining a large advantage.
Over the next few months, we are likely to see a major step forward again. The pace of Ai innovation means that it all comes down to one thing: speed.
