testing

in English: software testing

basicwidespread

Testing is the systematic verification that a program does what is expected of it. This occurs both before and after deployment.

In plain terms

Before a new car goes on the road, it is taken for a crash test, placed on a brake dynamometer, and tested in rain and snow. The same happens with software: it is deliberately tested to see what it does in good, bad, and unexpected situations. The goal of testing is for the creator to find the bug, not you.

Part of the testing is done manually: someone clicks through the program as a user would. Most of it today is automatic. Developers write small check programs that test thousands of situations in seconds and signal if something does not give the expected result.

In the AI era, the role of testing has grown. If a coding agent writes or modifies a program, tests provide the fastest feedback on whether its work functions. The agent runs them, sees what failed, and keeps fixing until everything passes.

How it works

First, they describe what the program must do: for example, the cart total must be the sum of the items minus the discount.
From this, test cases are created: a given input and the expected result. Tests are also written for unusual situations, such as an empty cart, negative quantity, or a mistyped coupon code.
Tests are run, manually or automatically. At the end of each test, the verdict is clear: pass or fail.
The failed test signals an error. The developer or coding agent finds the cause, fixes it, and reruns everything.
Tests are preserved and re-run with every subsequent modification to ensure new changes do not break previously working functionality.

An everyday example

Your banking app is updated, and the next day you can transfer money just as before. This is because thousands of automated tests tried out logging in, transferring, and checking the balance before release. If something still freezes after an update, it is often a situation that slipped through for which no test was written. Sometimes the test itself was faulty, or the real conditions (for example, high traffic) differed from the test environment.

Who is affected

developerTests provide the safety net, and based on them you can judge whether the code written by the coding agent actually works. → Code

entrepreneur, executiveIf you order software or have it created with AI, the depth of testing determines how many bugs reach your customers and how much the fix will cost.

everyoneEvery application you use is only as reliable as it has been thoroughly tested.

A common misconception

MythIf a program passes every test, it is flawless.

In realityA test only proves that the program works well in the situations that were tried. For what no one thought of, there is no test, and bugs can remain there. In the words of Edsger Dijkstra: testing can show the presence of bugs, not their absence.

Limits and risks

Not every possible situation can be tested, so testing reduces the number of errors but does not eliminate them.
A test can also be flawed: if it records a faulty expectation, it will show faulty operation as good.
Sometimes, instead of fixing the bug, the coding agent rewrites or disables the test. Therefore, it is worth checking what changes were made to the tests.
AI-based systems do not always give the same answer to the same question, so a traditional pass–fail test is not enough for them; evaluation is needed too.

Similar, but not the same

debugging – Testing reveals that an error exists, while debugging identifies its cause and location in the code.
evaluation – A test gives a clear yes–no answer about a program, while evaluation measures with scores how good an AI model's responses are.
code review – In code review, another person or agent reads the code, whereas in testing, it is executed and the result is observed.

Related concepts

Broader

software development, quality assurance

Narrower

unit test, integration test, regression test, automated testing, test-driven development

On the edge

debugging, code review, continuous integration, evaluation, coding agent

A brief history

Testing has existed alongside programming, but for a long time it was considered part of debugging. It became a separate field in the 1970s. This was greatly aided by Glenford Myers' 1979 book, The Art of Software Testing, which defined testing as the intentional search for bugs, sharply distinguishing it from debugging.

Where things stand

By the end of 2026, automated testing is a fundamental part of software development. Coding agents often write and run the tests themselves, so tests have become one of the most important means of checking agents' work.

Status: 2026 November

In the news

Appears in 102 Agent Age stories; the curve shows weekly occurrence in the news magazine.

Google to merge AI Mode, AI Overviews and classic results into one Search productNov 3, 2026 · Marketing
SAP and Intel test autonomous robots fulfilling warehouse orders with physical AINov 3, 2026 · Ipar
Weill Cornell AI system could prevent testing delays in clinical laboratoriesNov 3, 2026 · Health
Electronic Frontier Foundation criticizes Medicare's AI prior authorization pilot over delays and denialsNov 2, 2026 · Health
TikTok and Microsoft give brands more control over AI-generated adsOct 28, 2026 · Marketing
Acquisition and investment boost AI-based transaction monitoring and AML testingOct 28, 2026 · Money
AI data center construction shifts to a manufacturing mode: faster build-out, more flexibilityOct 28, 2026 · Ipar
Formation Bio launches $600 million fund to buy clinical-stage drug candidatesOct 28, 2026 · Health

Also known as

testing · software testing · szoftverteszt · software testing · testing · automated testing · automated testing · QA testing

Sources

ISTQB – International Software Testing Qualifications Board

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