Test what AI wrote

AI can write tests as quickly as code, and that creates a trap. A test written from the code can simply repeat the code's mistakes, and then a wrong answer shows up as a green tick.

// This proves nothing: the real function is replaced by a fake
jest.mock("./price", () => ({ applyDiscount: () => 900 }));

test("10% off 1000 is 900", () => {
  expect(applyDiscount(1000, 10)).toBe(900);
});

Better habits

• Write down what should happen before you look at the code.
• Test the edges: zero, empty, negative, and too large.
• Test what must be refused, not only what must work.
• Break the code on purpose. If the test stays green, the test is useless.

test("a discount over 100% is refused", () => {
  expect(() => applyDiscount(1000, 110)).toThrow();
});

🔬 Try it: all green, nothing tested →