O
AI-generated image

adversarial

/ˌædvərˈsɛəriəl/ adjective · British & US
Valid in UKValid in US
Share WhatsApp

What does adversarial mean?

Adversarial is an adjective describing situations, relationships, or processes marked by opposition and conflict between two sides. It is most familiar from the phrase 'adversarial system', used to describe legal frameworks — such as those in the United States and United Kingdom — in which competing parties argue their cases before a neutral judge or jury, on the theory that truth emerges from open contest. Beyond law, the word appears in political science, journalism, and everyday descriptions of hostile negotiations or rival organizations. Its tone is formal to neutral: calling something adversarial suggests more than simple disagreement, implying entrenched, often zero-sum opposition. Since the 2010s, the word has also become central to artificial intelligence, where 'adversarial examples' are inputs deliberately crafted to fool machine-learning models, and 'generative adversarial networks' pit two algorithms against each other to produce strikingly realistic images. Derived from the Latin 'adversarius', meaning opponent, the word ultimately traces back to 'adversus', 'turned against'.

adjective

Involving or characterized by opposition, conflict, or hostility between two parties, especially in contexts such as law, politics, or games where one side acts against another.

Senses
  1. Relating to or characterized by conflict, opposition, or hostility between parties (e.g. adversarial legal systems, adversarial journalism).
  2. (Machine learning) Describing techniques in which models are trained or tested against opposing agents, as in generative adversarial networks (GANs) or adversarial examples that fool classifiers.
Example

"The two lawyers took an adversarial stance from the opening statements, each determined to dismantle the other's case."

More examples

"The country's adversarial legal system assumes that truth emerges best when opposing sides argue freely."

"Researchers used adversarial training to make the image classifier more robust against manipulated inputs."

Did you know?

The word 'adversarial' comes from the Latin 'adversarius', literally 'turned against you' — and in machine learning today, 'adversarial attacks' are crafted inputs designed to trick AI models into making mistakes.

Reviewed by Deb Chak, Editor. AI-assisted content curated by RJS Tech Solutions LLP.

Etymology of adversarial

Adversarial derives from the Latin 'adversarius', meaning 'opponent' or 'enemy', itself formed from 'adversus', 'turned against' or 'facing', the past participle of 'adverti', 'to turn toward'. The word entered English via the noun 'adversary', with the adjective form developing later, becoming well established by the nineteenth century in legal and political discourse. It shares this Latin root with related words such as 'adversity', 'adverse', and 'advertisement', all built on the same notion of turning or being turned against something.

Related word forms

How adversarial is actually used

Adversarial is a formal-to-neutral term common in legal, political, journalistic, and technical writing. It carries a connotation of structured or institutionalized conflict rather than mere rudeness; describing a relationship as adversarial often implies distrust or zero-sum dynamics. In recent decades it has gained a prominent second life in computer science, where 'adversarial' describes both defensive training methods and deliberate attempts to deceive AI systems.

Easily confused with adversarial

adversary

An adversary is the noun meaning an opponent or enemy, while adversarial is the adjective describing a relationship or attitude of opposition.

averse

Averse means feeling disinclined toward something, whereas adversarial describes active opposition between parties.

What's another word for adversarial?

Words and phrases paired with adversarial

adversarial relationshipadversarial systemadversarial processadversarial attack

What's the opposite of adversarial?

Rhymes with adversarial