This year, various industries across many organizations are implementing artificial intelligence. Often, leaders who are not familiar with Artificial Intelligence, Machine Learning, and Deep Learning will use the terms interchangeably. When working with technologists and AI experts, this can confuse them. In this post, I’m going to explain each of these terms.
Artificial Intelligence is when machines demonstrate human intelligence. In 2020, most of Artificial Intelligence deployed in commercial settings is still “Narrow AI.” These are machines that exhibit a very specific type of human intelligence and are used to perform very specific tasks. For instance, an NLP algorithm can detect emerging trends. We are still far away from another type of AI: artificial general intelligence (AGI). AGI is when AI is intelligent enough to learn as humans do.
Machine Learning is the practice of using algorithms to learn from data. It is a type of Artificial Intelligence that is widely used today. Many algorithms were developed in the early days of AI that are still used today: clustering, random forest, reinforcement learning, bayesian networks, and more. These algorithms train on large sets of data that enables the algorithm to learn how to perform a specific task. As the algorithm learns from more data, it can get progressively accurate in its predictions.
Applications of machine learning algorithms can be found in computer vision, spam filtering, automated customer support, product recommendations, and more.
There are three types of machine learning approaches supervised learning (using labeled data), unsupervised learning (using un-labeled data), and reinforcement learning (learning to maximize rewards). Each of these algorithms has different use-cases. Often a combination of supervised learning and unsupervised learning is used in the real world depending on the type of data available and the type of tasks it’s trying to duplicate.
Under the scope of machine learning, there are different types of algorithms. As early classification algorithms became insufficient, researchers developed deep learning algorithm modeling after the mechanism inside the human brain. Specifically, artificial neural networks are the foundation of deep learning algorithms. Like the human brain, connections are made between different layers of neural networks where data can propagate across. Weightings of neurons are tuned precisely to generate the right answers each time.
Deep learning is a sub-field of machine learning concerned with algorithms inspired by the structure and function of the brain called artificial neural networks. Algorithms are often computationally intensive. They break down tasks in ways that enable AI-enabled applications. Andrew Ng was one of the pioneers of deep learning. He implemented artificial neural networks across millions of YouTube videos while working at Google. Today, deep learning applications can be found in self-driving cars, tumor identification, and recommendation systems.
Even with deep learning’s wide applications, we are still focused on Narrow AI. As artificial intelligence progress, we will likely see more advanced AI algorithms to complement the usage of deep learning. Even in the scope of Narrow AI, variations of deep learning algorithms may emerge to use fewer data and with improved performance.