Exponential Growth - What it means for artificial intelligence and the future
Now that you're here, we can begin.
For most people, I suspect the phrase, “exponential growth,” probably doesn’t generate much excitement. But once you’ve read this post, that will probably change. That is, if you’re anything like me or Ray Kurzweil, Peter Diamandis and others (to be clear, I am not placing myself among those incredibly accomplished individuals).
Humans think linearly. If it took 5 years to build half of a bridge, we naturally think it'll take another 5 years to complete it.
That's essentially how linear growth works. By “linear growth” I mean something that grows or advances by the same amount each step, regardless of the total underlying amount of progress or quantity.
For example, if you are paid $5 per day, your income is growing linearly, i.e., by $5 each step. It doesn't matter if you start with $5 or $10,000, the amount will always increase by $5:
Easy, right? Ok, let's use another example. The first working mobile phone debuted in 1973. Fast forward approximately 34 years to 2007, Apple introduces the iPhone (yes, I know Apple didn’t invent the smartphone).
Thinking linearly about this: since it took roughly 34 years to go from the first working mobile phone to the iPhone, then it will take approximately 34 years for the next generation of smartphones to make the original iPhone look like the original mobile phone. Whew, what a sentence.
This process, linear growth, is something that we’re all used to. And it applies to so many facets of our lives that we tend to think of nearly everything in a linear fashion:
But exponential growth is very different. It is the growth of a system in which the amount being added to the system is proportional to the amount already present: the bigger the system is, the greater the increase.
So let's look at our $5 a day example. If your pay grew exponentially, doubling each day, this is what it would look like:
Big difference right? We're just getting warmed up.
If you take nothing else from this post, remember this: the speed of exponential growth is deceiving. This is because linear growth and exponential growth look very similar in the beginning. Then, very suddenly, they diverge.
I call this, the "whiplash effect."
The Whiplash Effect - All you need is the AI 9000 and Intel Blocks
You and I are brilliant scientists (just work with me here). We're tirelessly working on creating artificial super intelligence. Humanity will change forever.
After many years, we've just completed our magical machine, the AI 9000.
That's right. The AI NINE-FUCKING-THOUSAND.
The AI 9000 builds and assembles "Intel Blocks." Intel Blocks are obviously the units that, when enough are assembled, form artificial super intelligence (duh). They're kind of like artificial super intelligence Legos.
Unbeknownst to us, it will take the AI 9000 30 years to assemble enough Intel Blocks to create artificial super intelligence. The AI 9000 assembles the Intel Blocks at an exponential rate, doubling the number of Intel Blocks each year.
Here's the thing, we've been asked by all world leaders to give them a heads-up on the progress of the AI 9000. You know, because the world might irreversibly change once this artificial super intelligence thing is running around in the wild.
We decide we'll let the world leaders know once the AI 9000 is half-way finished. That should give them plenty of time, right? So we open the AI 9000 app and set the "Remind Me to Alert All World Leaders" at 50% Intel Blocks. We then forget about it and go on with our other work.
Years later, our "Remind Me to Alert All World Leaders" alarm goes off.
"We've hit 50% of required Intel Blocks," we say with excitement. "We're half-way there!"
So we call the world leaders and let them know there's nothing to worry about. The AI 9000 still has another 50% to go.
"It's only at the half-way point," we say, "so you have many more years before anyone needs to worry about this artificial super intelligence business."
"Great," say the world leaders in unison. Then they hang up without saying "good-bye" like they do in the movies (which is inexplicable by the way, no one ends a phone conversation like that).
We go back to work. The AI 9000 keeps humming along.
One year passes. Year 30 has arrived.
We're working in our lab when suddenly the AI 9000 begins smoking. Then, much to our horror, the AI 9000 displays the "Holy Shit, the Artificial Super Intelligence is Complete" status message.
This is serious.
I mean, just look at the AI 9000.
We look at each other in disbelief.
"That can't be, the AI 9000 was only half-way finished two years ago! The AI 9000 can't be finished!", my voice echoes throughout the lab.
You reply, "I think you're mistaken. If you remember, the AI 9000 displayed the 'Holy Shit, the Artificial Super Intelligence is Complete' status message. Like just a few seconds ago." Your input was very helpful.
We call the world leaders, who then have us arrested.
No there aren't any plot holes here.
Ok, the literary brilliance you just suffered through has a purpose. It took the AI 9000 28 years to build 50% of the necessary Intel Blocks. It then built the rest of the Intel Blocks only two years later.
Think about this: the AI 9000 assembled 28 years worth of Intel Blocks in the final year. IN. TWO. YEARS.
This, my friends, is the whiplash effect. In the beginning, exponential growth looks very similar to linear growth. Then, quite suddenly, it skyrockets.
So let's recap. Linear growth looks like this:
Exponential growth looks like this:
Now look at linear growth and exponential growth together:
You'll notice the two lines are nearly identical at the beginning. After all, in our $5 a day example, both charts show we made the same amount of money on Monday ($5) and Tuesday ($10). But once we got to Wednesday, the linear and exponential charts diverged.
Ok, ok. You get it. So why is this important?
To quote Dr. Thomas Lovejoy of the Smithsonian Institution, “A lack of appreciation for what exponential increase really means leads society to be disastrously sluggish in acting on critical issues.”
Which is exactly what we'll look at next.
The Exponential Advancement of Technology
Have you ever heard of Ray Kurzweil? If not, here's a little background.
Ray Kurzweil is currently the Director of Engineering at Google where he works "on new projects involving machine learning and language processing". Kurzweil, an MIT graduate, was the principal inventor of the first charge-coupled device flatbed scanner, the first omni-font optical character recognition, the first print-to-speech reading machine for the blind, the first commercial text-to-speech synthesizer, the Kurzweil K250 music synthesizer capable of simulating the sound of the grand piano and other orchestral instruments, and the first commercially marketed large-vocabulary speech recognition.
Kurzweil has also written several books: The Singularity is Near: When Humans Transcend Biology, How to Create a Mind: The Secret of Human Thought Revealed, and The Age of Spiritual Machines: When Computers Exceed Human Intelligence.
So yeah, he's kind of a smart guy.
Ray Kurzweil also came up with this idea called, "the law of accelerating returns." He explains it better than I ever could:
An analysis of the history of technology shows that technological change is exponential, contrary to the common-sense “intuitive linear” view. So we won’t experience 100 years of progress in the 21st century — it will be more like 20,000 years of progress (at today’s rate). The “returns,” such as chip speed and cost-effectiveness, also increase exponentially. There’s even exponential growth in the rate of exponential growth. Within a few decades, machine intelligence will surpass human intelligence, leading to The Singularity — technological change so rapid and profound it represents a rupture in the fabric of human history. The implications include the merger of biological and nonbiological intelligence, immortal software-based humans, and ultra-high levels of intelligence that expand outward in the universe at the speed of light.
A lot of people misunderstand Kurzweil and merely assume that he bases the law of accelerating returns on Moore's law.
Moore's law is the observation that the number of transistors in a dense integrated circuit doubles approximately every two years. The observation is named after Gordon Moore, the co-founder of Fairchild Semiconductor and Intel, whose 1965 paper described a doubling every year in the number of components per integrated circuit, and projected this rate of growth would continue for at least another decade.
Moore's law didn't conclude in a decade, it hasn't stopped. But it will soon. In fact, Kurzweil predicts Moore's law "will die a dignified death no later than the year 2019."
Many people quote Moore's law as the foundation for the exponential advancement of technology. But they really shouldn't.
The first reason is, physical limitations will eventually halt the production of traditional integrated circuits (the size of the atom is a hard-stop to making smaller circuitry). So Moore's law isn't all that helpful for projecting exponential growth of technology past the next decade.
The second reason is, the law of accelerating returns isn't based on Moore's law. It's an observation that, throughout history, technological advancement has been growing exponentially across an array of different technologies. As Kurzweil puts it:
One can examine the data in different ways, on different time scales, and for a wide variety of technologies ranging from electronic to biological, and the acceleration of progress and growth applies. Indeed, we find not just simple exponential growth, but “double” exponential growth, meaning that the rate of exponential growth is itself growing exponentially. These observations do not rely merely on an assumption of the continuation of Moore’s law (i.e., the exponential shrinking of transistor sizes on an integrated circuit), but is based on a rich model of diverse technological processes. What it clearly shows is that technology, particularly the pace of technological change, advances (at least) exponentially, not linearly, and has been doing so since the advent of technology, indeed since the advent of evolution on Earth.
So will the exponential growth of technology continue? I believe it will.
Humans innovate in the face of problems. And when you consider how much the world relies on technology and the entire industries built upon it, the biggest technological problems are receiving the largest amount of resources, both financial and intellectual.
And according to Ray Kurzweil, "the exponential growth of computing didn’t start with integrated circuits (around 1958), or even transistors (around 1947), but goes back to the electromechanical calculators used in the 1890 and 1900 U.S. Census." He points out that there have been "at least five distinct paradigms of computing, of which Moore’s law pertains to only the latest one."
As an example of where we might go from here with computing, I would encourage you to watch the incredibly interesting TED Talk by Dr. George Tulevski. Dr. George Tulevski received his Ph.D. in Chemistry from Columbia University in 2006. He was hired as a Research Staff Member at IBM’s TJ Watson Research Laboratory in 2008 where he currently works in the carbon nanoelectronics research group. In his TED Talk, he discusses where the computing world may potentially go once quantum mechanics halt the production of traditional integrated circuits along the Moore's law timeline.
As Dr. Tulevski mentions, there was little incentive to innovate when Moore's law ensured better performance and more efficiency with each coming generation of chips. But that's changed now. And, as a result, the gears of innovation have begun turning once again.
Here are a few examples of technologies that may be the answer for post-Moore's law computing from The Economist's outstanding article, After Moore's law:
Artificial Intelligence and the Future
If you believe it's more likely than not that technology will continue to advance at an exponential rate, then read on. If not, then read on.
Life as we know it will undergo profound changes over the next two to three decades. And even more so after that. Let me explain why.
It's difficult to define intelligence, but a pretty good definition was coined by Howard Gardner:
To my mind, a human intellectual competence must entail a set of skills of problem solving — enabling the individual to resolve genuine problems or difficulties that he or she encounters and, when appropriate, to create an effective product — and must also entail the potential for finding or creating problems — and thereby laying the groundwork for the acquisition of new knowledge.
The part I want to focus on is the creation of "an effective product...thereby laying the groundwork for the acquisition of new knowledge." All technology is a product of human intelligence. From the most basic tools to the most complex machines or software, all of it was created by humans to address a problem or aspect of life. In its most basic sense, human intelligence is a solution-maker.
For centuries, our tools have become more complex and sophisticated as a result of the application of human intelligence compounded over time. But no matter how complex or sophisticated, all of these advancements were a form of solution.
But now, we're talking about the invention of something that isn't just a solution, but a solution-maker itself: artificial intelligence. And artificial intelligence is inextricably intertwined with an exponentially advancing technology: computing.
In Pedros Domingo's fantastic book,The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World, he describes the point where machine intelligence exceeds human intelligence as the "Turing Point." Referring to our trajectory as a "phase transition," he states the Turing Point "will come when machine learning overtakes the natural variety. Natural learning itself has gone through three phases: evolution, the brain, and culture. Each is a product of the previous one, and each learns faster. Machine learning is the logical next stage of the progression."
This is why, for the first time in human history, the fundamental assumptions of our existence are going to change in the relatively near future, i.e., in the next 20 to 30 years.
Almost everyone reading this is likely part of the most fortunate group of people in our collective history. We will see the world change in ways that are unimaginable. Not 100 years from now, but 20 to 30 years at the most. This is because of exponential growth.
As stated by Gideon Lewis-Kraus in his New York Times piece, The Great A.I. Awakening:
[The Google Brain team doesn't] assume that “consciousness” is some special, numinously glowing mental attribute — what the philosopher Gilbert Ryle called the “ghost in the machine.” They just believe instead that the complex assortment of skills we call “consciousness” has randomly emerged from the coordinated activity of many different simple mechanisms. The implication is that our facility with what we consider the higher registers of thought are no different in kind from what we’re tempted to perceive as the lower registers. Logical reasoning, on this account, is seen as a lucky adaptation; so is the ability to throw and catch a ball. Artificial intelligence is not about building a mind; it’s about the improvement of tools to solve problems.
As processes are developed to tackle more and more problems, both simple and difficult, the aggregate effect begins to coalesce into something different. This will take many by surprise because they think in terms of linear growth, not exponential growth.
This gif from Medium demonstrates this point rather well:
Let's take a closer look:
The point is, we aren't far off. Advanced artificial intelligence isn't hundreds of years away.
And it's not just Ray Kurzweil that believes this. Experts in the field of artificial intelligence believe there is a 50% chance that "high level machine intelligence will be developed around 2040-2050."
That's why exponential growth in this context is so important. All of us need to think in exponential terms. And we need to begin thinking about the impacts of these technologies.
But there is something else to think about as well. And it's a lot more fun, at least to me.
Take a moment to reflect that you are likely going to see the world change more than it ever has before. The concept of life, humanity's place in the universe, the nature of our existence - all of it is on the table.
An exponentially advancing solution-maker can be applied to everything: genetic engineering, materials sciences, energy, physics, cancer treatment, anti-aging - the list goes on and on.
And we are going to get to see it happen. What could be more exciting?