Smart & Connected Life Smart Home What Is Machine Learning? How does this application of AI work? by Robert Earl Wells III Writer Robert Wells is a professional writer and amateur game developer. His specialties include web development, cryptocurrency, and cybersecurity. our editorial process LinkedIn Robert Earl Wells III Updated on December 17, 2019 Smart Home Your Best Year Ever: College Tech Tips Amazon Appliances & Lighting Google Tweet Share Email Machine learning allows computer programs to execute new tasks without explicit instructions from developers. Here's what machine learning is, how it relates to artificial intelligence (AI), and examples of its wide-ranging applications. What Is Machine Learning? Machines have traditionally relied on algorithms, or sets of instructions, to execute specific tasks. Machine learning involves computers building upon past experiences to make predictions and formulate new solutions to problems with minimal human input. For example, social media websites like YouTube use machine learning algorithms to curate content based on your personal interests. How Machine Learning Works Like humans, machines learn from experience. Also like humans, they learn in different ways. Some machine learning algorithms rely on trial and error, while others draw inferences based on pattern recognition. Many AI agents use several types of learning algorithms to make decisions and devise plans to achieve goals. One example of machine learning that people rely on every day is voice assistants like Alexa and Google Assistant. These virtual assistants use a technique called natural language processing, or NLP, to understand and respond to voice commands in hundreds of languages. Computers can even create their own languages using the same technology. Whenever you talk to a virtual assistant, it uses the data from your interaction to improve its own voice recognition capabilities. That information is uploaded to the internet and shared for the benefit of all users. Big Data and Machine Learning Machine learning is nothing new, but the rise of big data has contributed to recent advancements in AI. Now that companies can collect and analyze massive amounts of data in the cloud, it's possible to create highly personalized user experiences. For example, streaming services like Netflix use machine learning algorithms to recommend movies and TV shows based on your viewing history. What Are Applications of Machine Learning? Other examples and applications of machine learning include: Banks: Banks use machine learning algorithms to detect fraud based on your spending patterns.Customer Service: Many companies use smart assistants to provide online customer service.Autonomous cars: Self-driving cars use various machine learning techniques to navigate the roads.Scientific research: Machine learning algorithms can be used to conduct rapid scientific testing for the development of medical treatments.Games: Computers have been trained to master board games like chess and even video games like Dota 2. Is Machine Learning AI? The definition of artificial intelligence has shifted over the decades. Artificial intelligence can be understood as machines performing tasks that only humans can perform, so once a machine can do something, it no longer counts as intelligence. This phenomenon is called the AI effect. Computer scientist Douglas Hofstadter famously observed that "AI is whatever hasn't been done yet." However, in recent years, rapid advancements in machine learning have resulted in computers that can outperform humans in increasingly complex cognitive tasks. So far, machine learning has been limited to specific goals predetermined by humans, but a computer capable of everything a human can do might one day become a reality. The Limitations of Machine Learning While machines that can reprogram themselves are far in the future, it's worth considering the social implications of machine learning and AI in the real world. Like people, computers have inherent biases based on previous experiences, so they're just as capable of discrimination. Furthermore, programmers who design machine learning algorithms must be aware of their own biases. Teaching machines morality will be significantly more challenging than teaching them mathematical reasoning.