Machine Learning

… a core concept used in Managing New Technologies and Atlas112M

Concept description

The National Research Council of Canada (reference below) defines machine learning as “the ability of computers to identify patterns, learn from data, and make inferences or decisions, without having been explicitly programmed to do so.”

It goes on to say:

“This exciting field is a key part of artificial intelligence, driving innovation in research methods, industrial operations, and consumer products like mobile devices and smart homes.

“The NRC excels at machine learning to query large volumes of text, perform machine translation and evaluation, uncover new molecular interactions, monitor engineering systems, and capture data through machine vision.”

In its glossary of basic artificial intelligence terms (reference below), TNW states:

“The meat and potatoes of AI is machine learning — in fact it’s typically acceptable to substitute the terms artificial intelligence and machine learning for one another. They aren’t quite the same, however, but connected.

“Machine learning is the process by which an AI uses algorithms to perform artificial intelligence functions. It’s the result of applying rules to create outcomes through an AI.”

Algorithms and machine learning

In their 2019 book, The Ethical Algorithm (reference below), Kearns and Roth describe the difference between human-designed algorithms and machine-designed algorithms:

“Algorithms such as the sorting algorithm … are typically coded by the scientists and engineers who design them: every step of the procedure is explicitly specified by its human designers, and written down in a general-purpose programming language such as Python or C++. But not all algorithms are like this. More complicated algorithms – the type that we categorize as machine learning algorithms – are automatically derived from data. A human being might hand-code the process (or meta-algorithm) by which the final algorithm – sometimes called a model – is derived from the data, but she doesn’t directly design the model itself.

“In traditional algorithm design, while the output might be useful (like a sorted list of Facebook usage times, which could help in analyzing the demographic properties of the most engaged users), that output is not itself another algorithm that can be directly applied to further data. In contrast, in machine learning, that’s the entire point. For example, think about taking a database of high school information about previously admitted college students, some of whom graduated from college and some of whom did not, and using it to derive a model predicting the likelihood of graduation for future applicants. Rather than trying to directly specify an algorithm for making these predictions – which could be quite difficult and subtle – we write a meta-algorithm that uses the historical data to derive our model or prediction algorithm. Machine learning is sometimes considered a form of “self-programming,” since it’s primarily the data that determines the detailed form of the learned model.

“This data-driven process is how we get algorithms for more human-like tasks, such as face recognition, language translation, and lots of other prediction problems that we’ll talk about in this book. Indeed, with the aforementioned explosion of consumer data enabled by the Internet, the machine learning approach to algorithm design is now much more the rule than the exception. But the less directly involved humans are with the final algorithm or model, the less aware they may be of the unintended ethical, moral, and other side effects of those models, which are the focus of this book.”

See also: Artificial Intelligence and Public Policy and The Ethical Algorithm – The Science of Socially Aware Algorithm Design (Michael Kearns and Aaron Roth, 2019).

Atlas topic, subject, and course

Managing New Technologies (core topic) in Information and Technology Management and Atlas112M Management of Human, Information, and Technology Resources.


National Research Council Canada (2020), Machine Learning, at, accessed 1 March 2021.

TNW (The Next Web, 2017), A glossary of basic artificial intelligence terms and concepts, at, accessed 1 March 2021.

Michael Kearns and Aaron Roth (2019), The Ethical Algorithm, Oxford University Press, Kindle Edition, pages 6-7.

Page created by: Ian Clark, last modified 1 March 2021.

Image: The Wordstream Blog, 10 Companies Using Machine Learning in Cool Ways, at, accessed 1 March 2021.