Machine Learning
Noun · AI & Machine Learning · Origin: 1959
Definitions
Machine learning is a subset of artificial intelligence in which systems learn patterns from data rather than being explicitly programmed with rules. Instead of writing code that specifies every decision, developers train models on datasets, and the models learn to make predictions, classify inputs, or generate outputs based on statistical patterns they discover. Machine learning is broadly categorized into supervised learning (training on labeled data to predict outcomes), unsupervised learning (finding hidden patterns in unlabeled data), and reinforcement learning (learning through trial and error with rewards). Common algorithms include linear regression, decision trees, random forests, support vector machines, neural networks, and transformer architectures. Deep learning, a subset using multi-layered neural networks, powers breakthroughs in image recognition, natural language processing, speech synthesis, and generative AI. Machine learning requires careful attention to data quality, feature engineering, model selection, hyperparameter tuning, and evaluation metrics. Popular frameworks include scikit-learn, TensorFlow, PyTorch, and XGBoost. Machine learning is transforming industries from healthcare and finance to autonomous vehicles and content recommendation.
In plain English: Teaching computers to learn from data instead of being given explicit instructions — the computer figures out the patterns on its own.
Etymology
- 1943
- Warren McCulloch and Walter Pitts publish a mathematical model of neural networks, laying groundwork for computational learning.
- 1959
- Arthur Samuel at IBM coins 'machine learning' while developing a checkers-playing program that improves through self-play.
- 1986
- Backpropagation gains traction through work by Rumelhart, Hinton, and Williams, reviving interest in neural networks.
- 2012
- AlexNet wins ImageNet by a wide margin, igniting the deep learning revolution.
- 2020s
- Large language models and generative AI bring machine learning into mainstream consciousness and everyday tools.
Origin Story
The Checkers Game That Taught Computers to Learn
In 1959, Arthur Samuel, an IBM engineer and former University of Illinois professor, published a landmark paper describing a checkers-playing program that could improve its own performance through experience. He coined the phrase 'machine learning' to describe this process, defining it as giving 'computers the ability to learn without being explicitly programmed.' Samuel's checkers program was remarkable: it studied thousands of board positions and developed its own evaluation strategies, eventually beating Samuel himself. The program ran on an IBM 701, a vacuum-tube mainframe that filled an entire room yet had less computing power than a modern pocket calculator. Samuel chose the name carefully. He wanted something that conveyed the fundamental shift from traditional programming, where humans dictate every step, to a new paradigm where the machine discovers patterns on its own. The field grew slowly through the 1960s and 1970s, suffered through the 'AI winters' of reduced funding and skepticism, then exploded in the 2010s when deep learning, massive datasets, and GPU computing converged to produce breakthroughs in image recognition, natural language processing, and game playing.
Coined by: Arthur Samuel
Context: IBM, 1959
Fun fact: Samuel's checkers program was so good that IBM initially worried about bad publicity, fearing the public would think computers were taking over.