In traditional software, a programmer writes the rules: if the transaction is over this amount and from a new country, flag it. In machine learning, you show a model many examples, such as transactions labeled fraud or not fraud, and a training process adjusts the model until its predictions match the examples. The rules end up encoded in the model’s learned parameters rather than written down anywhere.
Deep learning is the branch of machine learning that uses neural networks with many layers, and it has driven major advances in speech recognition, computer vision and language. Large language models are deep learning models trained on text. The practical consequence for anyone using them is the same as for any machine learning system: the model is only as good as the data it learned from, and it can be confidently wrong on cases unlike its training data.
Not every setup needs a labeled answer for each example. Supervised learning trains on examples with target answers, such as tickets paired with their correct queue. Unsupervised learning looks for structure, such as groups of similar tickets, without those labels. Reinforcement learning learns from rewards for actions (Google for Developers). Whichever setup is used, check the trained model on examples that were not used to fit or choose it, including ambiguous requests and changed wording.
Machine learning is one part of artificial intelligence, which also includes rule-based and search methods. Keeping a trained model useful after launch, with versioned data, testing and monitoring, is the work of MLOps.
Sources
- Google for Developers: What is Machine Learning? — Introduces supervised, unsupervised and reinforcement learning as different ways a model learns from data.
- LeCun, Bengio and Hinton: Deep learning — Reviews deep learning, models of many processing layers that learn representations from data, and its gains in speech recognition, object recognition and other domains; checked October 25, 2026.
Go deeper
From the conversation
-
Architecting AI Agents: The Shift from Models to Systems | Aishwarya Srinivasan -
The Accidental Algorithm | Humans of AI Crossover with Writer's Melisa Russak -
Mindset Over Metrics: How to Approach AI Engineering | Hamel Husain -
Gemini 3 & Robot Dogs: Inside Google DeepMind's AI Experiments | Paige Bailey -
AI's Two Extremes – Foundations & The Frontier | Databricks’ Denny Lee -
Your Key to AI Success is Hiding in Plain Sight | Cohesity's Greg Statton -
How Intercom Cut $250K/Month by Ditching GPT for Open Models | Fergal Reid -
Beyond Transformers: How Liquid AI Is Rethinking LLM Architecture | Maxime Labonne