This is a way for Microsoft to find and fix more than 30,000 errors generated by 47,000 programmers each month

Each month, 47,000 Microsoft programmers create more than 30,000 bugs. A number is too large to detect, Microsoft cannot use traditional methods such as tracking errors on GitHub and AzureDevOps repositories. So, how can the software giant detect and fix such a huge number of bugs?

Microsoft has created a machine-learning machine to help identify critical security errors that need to be fixed immediately with an accuracy of up to 99%. To create this 'engine', the technology giant had to use up to 20 years of data history, including 13 million jobs and errors.

Senior security director at Microsoft Scott Christiansen said their machine-learning system not only automatically detects errors, but can also divide them into extremely dangerous, serious or non-serious forms. , security or non-security with extreme accuracy. This helps Microsoft engineers know which bugs are extremely dangerous and need to focus on resolving them first.

In order to ensure that it is not outdated, Microsoft's machine-learning error detection system is still constantly being trained with new data.

This system helps the technology giant find 99% of the errors in huge quantities and accurately classify them to 97%.

1. Xbox 360 hardware error caused Microsoft to lose . 1 billion USD
2. New bug discovered on Microsoft Windows operating system