Google DeepMind (GDM)

basicwidespread

Google's artificial intelligence research and development division, which builds the Gemini models and scientific systems like AlphaFold.

In plain terms

Google DeepMind is Google's "AI laboratory". If Google is a car manufacturer, DeepMind is the engine development division. You do not necessarily see its name on the bodywork, but the models it develops power the AI features of many Google products, such as Search, Gmail, Android phones and the Gemini chatbot.

It does two things simultaneously. First, it conducts basic research: building systems that solve scientific problems, such as predicting the spatial structure of proteins or making weather forecasts more accurate. Second, it develops products: the Gemini language models, image and video generation models, and agents that perform tasks independently for you.

It differs from most competitors in that it is part of a giant corporation. It can rely on its own data centers, its own chips, and billions of users, but in return its decisions are also shaped by Google's business interests.

How it works

Researchers are developing new learning methods and models, and they publish some of the results in scientific journals.
The models are trained in Google data centers on massive amounts of text, images, audio, and video, primarily on the company's own chips (TPU).
Finished models undergo safety and quality tests, then are released in versions of different sizes: some are fast and cheap, others slower but more thorough.
Models are integrated into Google products (Search, Gemini app, Workspace, Android), and developers can also connect them to their own applications via programming interfaces.
Parts of scientific systems, such as the AlphaFold protein structure database, are freely available to researchers worldwide.

An everyday example

You enter a question into the Google search engine, and a summary answer appears above the results, or you ask Gemini on your phone to summarize a long email. In both cases, a model developed by Google DeepMind works in the background.

Who is affected

everyoneIf you use Google Search, Gmail, or an Android phone, DeepMind models often work in the background, even if you are unaware of it.

developerGemini models are accessible via a programming interface, allowing you to build your own applications and agents on them. → Code

patient, doctorAlphaFold has accelerated protein research, which could aid the development of new drugs and therapies in the long term. → Health

entrepreneur, executiveIf your company uses Google Workspace or Google Cloud, most of the available AI features are built on DeepMind models. Their development influences how you can use them.

A common misconception

MythDeepMind is an independent British research institute that focuses solely on games, chess, and Go.

In realityIt has been owned by Google since 2014 and, since 2023, has been the company's main AI research and development division under the name Google DeepMind. Games were only the early experimental ground; today, the Gemini models and scientific applications are at the center.

Limits and risks

Its models can also make mistakes and confidently state falsehoods, so always verify important information from other sources.
As part of a commercial enterprise, Google's business interests play a decisive role in what is free, what is paid, and what is discontinued.
Not every feature is available in every country and language: new features often arrive late in Hungary and may work less well in Hungarian.
Check the data protection settings, because depending on the type of service, the data you enter may be used to improve the systems.

Similar, but not the same

Gemini – Gemini is a product, meaning a model family and chat application, while Google DeepMind is the organization that creates it.
Google Brain – Google Brain was Google's former independent AI research group, which merged with DeepMind in 2023 to create Google DeepMind.
OpenAI – OpenAI is a separate company, the creator of ChatGPT, and a major competitor to Google DeepMind.

Related concepts

Broader

MI-labor, Google, Alphabet, Artificial intelligence

Narrower

Gemini, AlphaFold, AlphaGo, Gemma, Veo

On the edge

OpenAI, Anthropic, Meta AI, Large language model, Reinforcement learning

A brief history

DeepMind was founded in London in 2010 by Demis Hassabis, Shane Legg, and Mustafa Suleyman, and was acquired by Google in 2014. In 2016, its AlphaGo program defeated Lee Sedol (Hungarian transliteration: I Szedol), a South Korean Go grandmaster. In 2023, the company merged with the Google Brain group, and its name has since been Google DeepMind. For AlphaFold, Demis Hassabis and John Jumper received the 2024 Nobel Prize in Chemistry, which they shared with David Baker.

Where things stand

By the end of 2026, Google DeepMind is one of the world's leading AI labs. Gemini models are integrated into almost every Google product, and the lab is focusing increasingly on autonomous agents and systems supporting scientific research, alongside chat capabilities.

Status: 2026 November

In the news

Appears in 21 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Google's AI agent can propose new drug molecules within hours, DeepMind saysNov 3, 2026 · Health
Isomorphic Labs plans human trials soon of AI-designed cancer drugsOct 30, 2026 · Health
DeepMind unveils Genie 4 world model with more realistic physicsOct 25, 2026 · Agent Age
Google and DeepMind staff resign over company's military AI contractsOct 22, 2026 · Agent Age
Hassabis says AGI could arrive within two to three yearsOct 21, 2026 · Agent Age
Google adds Gemini-based scam protection to Android and SingaporeOct 14, 2026 · Money
Google releases Gemini 4 Argon; its agents completed Minecraft without human helpOct 13, 2026 · Agent Age
Microsoft's Brad Smith, Richard Hermer and 400 scientists call for tougher AI regulationOct 12, 2026 · Jog

Also known as

Google DeepMind · DeepMind · GDM · Deep Mind · DeepMind Technologies · Google Deepmind · a Google MI-laborja · Google's AI division

Sources

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.