Fri. Dec 6th, 2019

Artificial intelligence can now identify real authors by code style

1 min read

Programmers often have their styles, so the code written has its characteristics, which can be used to detect whether the code has plagiarism. Researchers have used this feature to create artificial intelligence applications, which only require fragment code to retrieve which programmer the code belongs to.

In the actual test, the accuracy of artificial intelligence recognition is more than 80%. In most cases, artificial intelligence can be identified in a short time. This is useful for detecting whether some code is plagiarized or plagiarizing someone else, but there are always two sides, and this technology also has privacy issues.

 

For example, programmers want to hide and usually create different code bases. Different account nicknames can confuse others to isolate their true identity. This situation is widespread in many open source software, and programmers often contribute code to an item anonymously without having to disclose their identity.

Switching accounts and nicknames under this artificial intelligence application are no longer anonymous, and as long as there is a public code snippet, the author can be successfully identified. Therefore, the author of this artificial intelligence application did not disclose its source code.

Via: wired