Material Insights · Understanding the Material World

Deep learning: Overview

What it does

The job the technology does for people, in a short paragraph

Deep learning lets computers learn tasks from examples rather than from rules written by people. It recognises objects in photos, turns speech into text and translates [1], and it runs in AI medical devices approved by regulators [4]. Its methods date from 1986 [1], it read bank cheques commercially by 1998 [7], and it took off in 2012 [1]. In 2024, 78% of survey respondents said their organisations used AI [4].

as of 2026-11-11

How it works

The working principle in plain words, step by step

A neural network is a stack of layers of simple units, loosely modelled on nerve cells. Each unit adds up its inputs, each multiplied by a weight, and passes on a signal if the total is high enough. The first layer takes raw data, such as the pixels of an image; the last gives an answer, such as "cat". In training, the network sees an example, makes a guess and measures how wrong it is. Backpropagation then works backwards through the layers and adjusts every weight slightly to make the error smaller [1]. Repeated over millions of examples, the hidden layers learn their own features: edges, then shapes, then whole objects. Convolutional networks, developed by LeCun in the 1980s, reuse the same small filters across an image [1].

as of 2026-11-11

Key figures

The handful of numbers that size it up, each with its unit, year and source

FigureValueUnitYearSource
Share of survey respondents whose organisations used AI78%2024[4]
FDA-approved, AI-enabled medical devices (cumulative)223devices2023[4]
US private investment in AI109billion USD2024[4]
AI servers' share of data-centre energy demand15%2024[6]
Notable AI models produced by US institutions40models2024[4]

as of 2026-11-11. The AI Index figures cover AI in general, not deep learning alone; the energy share is the IEA's estimate as reported by Scientific American.

Where it is used

The main places and products the technology is found in

  • Photo and video recognition, from phone cameras to medical scans
  • Speech recognition in phones and voice assistants
  • Translation and other language tasks
  • Driver assistance and robots
  • Fraud detection, recommendations and advertising

as of 2026-11-11

Scale today

How much of it there is now: units in use, made each year and who uses it

Users
78% of survey respondents said their organisations used AI in 2024 [4]
Unit
share of organisations
Year
2024
Note
No count of deployed deep learning models exists; organisational use of AI is the best available measure.

as of 2026-11-11

Sources

  1. Fathers of the Deep Learning Revolution Receive ACM A.M. Turing Award Association for Computing Machinery, news release distributed on EurekAlert!, 2019
  2. AI Index 2025: State of AI in 10 Charts Stanford Institute for Human-Centered AI (HAI), 2025
  3. Data centers will use twice as much energy by 2030, driven by AI (IEA Energy and AI) Scientific American, 2025
  4. Gradient-based Learning Applied to Document Recognition Proceedings of the IEEE (Yann LeCun, Leon Bottou, Yoshua Bengio and Patrick Haffner), article page, 1998