Artificial Intelligence, Machine Learning, and Self-Driving Cars

Artificial intelligence. Machine learning. Self-driving cars. If you’re keeping up with the rapid changes in the technology industry, you’re seeing a bunch of terms thrown around as if they’re interchangeable—but really, there are some pretty important distinctions. In this post, we’re going to demystify the differences, and clarify the relationships, among these terms, especially artificial intelligence, machine learning, and self-driving cars. Let’s begin with a simple model for how we’ll approach this topic:

Artificial intelligence is the ‘what’.

Machine learning is the ‘how’.

Self-driving cars are the ‘why’.

The What: Artificial Intelligence

Artificial intelligence is the broad field that covers all sorts of different initiatives and efforts to create machines that behave intelligently. What exactly it means to ‘behave intelligently’ is a question best left for the philosophers and cognitive scientists, but for us, it refers to creating machines that do the highly complex things that only humans have previously been able to do.

That means that AI is about creating machines that do more than just follow the commands that we give them. They can process input, make decisions, and take action. All programs can do that to some extent, but AI is defined by its complexity: the actions it takes aren’t simple executions of specific lines of code, but rather, they are decisions that emerge out of complex reasoning.

This is why we say artificial intelligence is the ‘what’. It’s what we want to create.

Knowing what we want to create, however, is only a fraction of the challenge before us. The real question is: how do we create it?

The How: Machine Learning

Machine learning is one answer to that question. Machine learning is one way we could create artificial intelligence. At its core, machine learning is a suite of complex statistical methods that can find the slightest trends and patterns in massive amounts of data thanks to the recent advances in cloud and high-performance computing. That’s ultimately what machine learning is about: pattern recognition. Pattern recognition is part of what makes humans intelligent: it’s part of what enables us to learn from experience or transfer knowledge from one domain to another. Machine learning gives that intelligence to machines.

What makes machine learning unique is its incredible power to process far more data than a single human could ever consume. A single AI agent powered by machine learning could read millions of medical scans and learn which patterns indicate a certain illness or condition. It could analyze billions of financial transactions and find clusters that suggest fraudulent activity. It could process decades of educational data and identify what interventions or programs are most successful at increasing long-term graduation rates.

Of course, machine learning is only one way to pursue artificial intelligence. There are numerous others that are less reliant on massive databases or incredible computing power. There are also others that are better equipped to act in unfamiliar situations or develop human-like understanding of concepts. The important takeaway here is that if artificial intelligence is what we want to create, machine learning is one answer to how we might create it.

So, artificial intelligence is what we want to create, and machine learning is one answer to how we create it. But to a certain extent, isn’t that putting the cart before the horse? Shouldn’t we be asking ourselves why we want to create AI in the first place?

The Why: Self-Driving Cars

Self-driving cars represent one of many answers to that question. Self-driving cars carry massive potential to solve enormous problems facing mankind today. An autonomous vehicle is effectively equipped with dozens of pairs of eyes, all connected to a brain wholly focused on safe and efficient driving, which is itself instantly connected to every other car on the road to learn from their experiences and communicate with them at the speed of light. When you combine that singular focus, that comprehensive array of sensors, and that fleet-level communication and learning, you have the potential to create the world’s safest, most efficient vehicle.

But to do that, you need to make it intelligent. You need to equip it with the ability to recognize pedestrians, stop lights, and street signs. Its communication with other cars needs to come with rules for decision-making and planning. It must understand the human context of navigation, like ‘dropping off’ and ‘following.’ It must be able to do hundreds of tasks that have previously been relegated to human intelligence; in other words, it must have artificial intelligence.

What, How, Why

Self-driving cars are one of many technologies with enormous potential to radically improve human health and well-being; that’s why we want to create them. Machine learning, with its ability to harness incredible computing power to create systems that can evaluate massive quantities of data and make decisions in real-time, is one answer to how we can make this a reality. And of course artificial intelligence itself is what we want to achieve—machines that can mimic human-level intelligence.

~

Artificial Intelligence Nanodegree Program

  (applications close November 14th!)

Machine Learning Engineer Nanodegree Program

Self-Driving Car Engineer Nanodegree Program

David Joyner
David Joyner
Eleven-year veteran of Georgia Tech, from undergrad to PhD to now delivering and teaching in the Online Masters of CS program. Passionate about using AI to deliver individualized, scalable educational experiences. Always looking for something new to learn.