Dr. Kaplan’s Blog #4: Artificial Intelligence, Machine Learning, and GAI

Artificial Intelligence (AI) has become a hot buzzword these days. Most of what people mean when they refer to AI, is actually machine learning. Machines are shown tons of data (the history of everywhere we click on the web, for example) so they can find and “learn” patterns in the data. These patterns then power intelligent action. For example, machines can “learn” to send us more targeted ads based on patterns in our clicks on the web.

What makes machine learning so powerful is that computers can learn and improve with little or no human intervention. Alpha Go – the program that beat the human world champion in the very complex game of “Go” – achieved its world class performance by playing itself millions of times. Computers never sleep, so they can learn 24X7. And, as chips become ever faster, computers can play millions of games in the time it would take a human to play a single game. Or they can drive millions of simulated miles in the time needed for a human to drive across town.

Because computer chips are vastly faster than human brains, we might expect AI to become vastly more intelligent than humans due to faster processing alone. But machines have another advantage. Our human brains must fit in our skulls, which in turn are sized to fit through the human birth canal. That means we are stuck with brains roughly the size of a Nerf football. Machines, on the other hand, have no size limit. Theoretically, an artificial brain can be the size of a fingernail, a football, a football field, or even a city. So, the machines are not only faster than humans, they can have vastly larger “brains” as well.

Why haven’t they surpassed humans -- not just in playing games, but in all intelligent activities? Why haven’t they already achieved General Artificial Intelligence (GAI) – sometimes also known as Artificial General Intelligence (AGI)?

The detailed answers to these questions are complex. The short answer is that computers are NOT YET smarter than humans at all things, but they WILL BE … and sooner than most people realize.

Today humanity is somewhere in between the rise of AI (described in Episode 4) and General Artificial Intelligence (described in Episode 5). Humans are at a critical period in the evolution of intelligent systems. We are still, arguably, the smartest organisms on the planet and the creators of the current generation of AI. But in the future, GAI will become vastly more intelligent than us. And GAI will ultimately be able to design future generations of GAI without any human involvement.

So we have this precious window of time to try to influence the development of GAI while humans are still needed. Our goal should be to ensure that GAI cherishes humanity instead of destroying us.

It is critical to get the word out to as many people as possible while humans are still able to influence the evolution of intelligent systems. That’s why I end every blog post with a request that you share the link with anyone who might have interest.

Thank You! - Craig