Artificial intelligence has proven effective at identifying security vulnerabilities, but early tests indicate it won’t be able to replace humans for a while.
