And AI Is Eating Software
We are going to explore how fast things truly are moving. This is going to affect everything, from our investments to our job prospects. Most areas of life will be completely upended.
The last 50 years saw a huge shift forward in our technology. A great deal of this was due to advancements made in computers, specifically semiconductors. Many are familiar with what was at the core of it.
Roughly 60 years ago, Gordon Moore made an observation that became known as **Moore's Law"". Basically it said that the number of transistors on a chip would double every two years. For most of the last 6 decades, that held true.
It also had enormous impact upon the path of humanity. This single "law" is what turned the information-computer generation into what it is today. Everything we do digitally is tied to the advancements made in this area.
The End Of Moore's Law
The challenge is that we are likely seeing the end of Moore's Law. We cannot keep the processors rates going. Things have gotten so small that is it getting physically impossible to pack more on there. Noise is becoming an issue. We are also seeing the cost and time to next level advancement increasing.
Hence other options are needed.
One of the biggest areas that is getting a lot of attention is changing the design. This will provide better optimization and use software to enhance the output of the devices.
We are also seeing the introduction of AI into the equation. In fact, many feel that chip verticalization could really enhance the productivity of these machines. This will cater to more rapid advancement.
Moore's Law of AI
Marc Andreeson said that "software is eating the world". This is something that many of us can latch onto. However, this is really evident in the following chart.
Whereas Moore's Law had a doubling every 2 years, AI software is doubling in capability every 3-4 months.
Notice the trajectory of Moore's Law. Then consider how far we have come with computation (and our devices) since the 1960s. Now look at the path that AI is on. Just consider what that will look like after a few decades.
It is easy to see how the chart will end up looking something like this one.
Source
This is what exponential progress looks like. In hindsight, what was viewed as rapidly progression is really tiny in comparison. When looking at the impact of Ai driven development compared to classical advancement, we see how quickly our world can be changing.
How much of an impact can we expect?
Automation has been part of chip design since the 1980s. But now AI-designed chips are producing great results that could lead to 1,000 times better performance for chips in the next decade, according to Synopsys CEO Aart de Geus.
We are seeing the shift from scaling to system complexity. This is looking at all aspects of design and optimizing them. Of course, humans are being moved up the scale since the legwork is being done by AI.
Applying AI to address complexity, power, and scaling requirements to reach the 1,000 times performance target is critical to moving beyond Moore’s Law, which previously relied on manufacturing advances for productivity improvements. As those manufacturing improvements hit limits because miniaturization is now at the atomic level, other kinds of improvements in design have to take their place.
Packing 40 Years Into 10
There is little dispute that the last 40 years saw a major leap in human progress. The power of the semiconductor completely changed our world. We saw these get smaller, less expensive, and more powerful.
Much of what we see around us can be traced back to this advancement. The improved communication systems that allow for inexpensive (free) global communication is evident from here. Satellite communications, smartphones, and the entire Internet only took form as Moore's Law keep forging ahead.
However, as we can see from the above chart, Moore's Law might be ending yet that is of no consequence. Because of new system designs driven by AI, something that is advancing at 8-9 times the pace of Moore's Law, we are going to see our computational ability advance more in 10 year than it did in the last 40.
What impact does this have on our different industries? How will this affect healthcare, construction, or robotics? Will this make things such as self-driving vehicles something that is mastered in 5 or so years? Many might say no way but looking at the curve we are on, is it wise to deny the possibility?
Does anyone think that jobs can stand up to the oncoming advancement in AI that we are witnessing here? Even if we get nowhere near AGI, the fact that our effective computational power will increase tremendously through better system design means that whatever we create will be orders of magnitude more powerful than today.
Just consider, for a second, that if the pace is accurate, that by the end of the year, AI capability will double again. Where we stand today is only half of what we will see by the close of 2021.
Automation is going to become commonplace. We will see AI trading systems. Office work is going to be increasing taken over by RPAs. Robotics will see improvement in their software capabilities while advancements in the material sciences, likely due to AI, could help with the hardware.
In short, everything is about to go parabolic. Do not believe for a second this cannot take place. Since the progress over the last 40 was due to advancement in computation, it stands to reason that this will continue.
We could be upon the pathway which is going to drive us forward.
If you found this article informative, please give an upvote and rehive.
gif by @doze
logo by @st8z