Using Reinforcement Learning to Write Java Unit Tests

You probably learned that hacking on code until it works is the wrong way to write a program. But that’s essentially what AI does (albeit trillions of times faster), to remarkable effect.

Diffblue’s CEO Mathew Lodge explains how AI can use reinforcement learning to do a probabilistic search of potential solutions when the search space is polynomial—such as when playing a game of Go against a grandmaster, or when writing regression unit tests.

Learn how we apply this technique at Diffblue to generate unit test programs for Java projects, hacking the code using reinforcement learning to find the tests that achieve coverage and usefulness goals. The resulting test suites find more problems earlier in DevOps pipelines.