From the course: Introduction to Artificial Intelligence

Searching for patterns in data

From the course: Introduction to Artificial Intelligence

Searching for patterns in data

“

- Over the last 30 years, machine learning systems have become the dominant form of artificial intelligence. That's because these systems are exceptionally well designed to look for patterns in massive data sets. Machine learning has also been supercharged with the wide availability of digital data. If you want to create an AI program to identify dogs, you now have access to millions of images. You can feed your network and help it learn with a large volume of available data. It's the same with other types of data. You can easily get digital video, audio, images and documents. Just a few decades ago, it would've been extremely difficult to get even a few thousand digital images. Now it's trivial to get access to all kinds of data. Remember that machine learning systems feed on data to learn new things. The more data you feed into the network, the easier it will be for the machine to identify patterns. Think about the system that you're using right now. This is a professional social network that provides video training, users watch the training through an online video player. That video player collects data about how often you fast forward or how long you watch before moving on to the next lesson. Now, suppose that the player records that data for everyone who watches the videos. That might be hundreds of thousands of videos and millions of users. So that's a lot of data. No human could look through all that data and gain any meaning from it. But machine learning algorithms look through this data and find patterns. You can see which content users find more interesting. This is exactly the type of data that many businesses have been looking for. You can now see real time patterns in how your customer interacts with your product. In many ways, this data can tell you a huge amount not just about your customer's interest but even broader trends in industries. This data has an enormous amount of value. You can use it to build new customer products or to improve products you already have. It's no coincidence that companies like Google and Microsoft are most enthusiastic about AI. In many ways, their whole business has been built on using machines to interpret massive data sets. This type of pattern matching can be a huge competitive advantage. Plus, newer artificial neural networks now allow machines to find patterns in even larger data sets. When just a few decades ago, these patterns would've been imperceptible with regular machine learning algorithms. In fact, one of the largest challenges around machine learning is that humans don't really know how the machine identifies these patterns. It's like a black box of data and processing power. People simply can't process data at that same level. So if your organization is starting its own AI program they should be comfortable with the fact that the network might be sensing things that humans aren't able to perceive. This might not be a challenge with most companies but might be a real problem with industries such as insurance and healthcare. You don't want these systems making decisions about your customer's health and safety that humans can't understand. Artificial intelligence is not the same as human intelligence, and even though we might reach the same conclusions, we're definitely not going through the same process. Think about the type of data your organization collects. Are they using the data for machine learning? If so, what kind of patterns is the machine identifying and for what purpose?

Contents