AI system analyzes code similarities, makes progress toward automated coding
With the rapid advances in artificial intelligence (AI), are we getting to the point when computers will be smart enough to write their own code and be done with human coders? New research suggests we might be getting closer to that milestone.
Researchers from MIT and Georgia Tech teamed with Intel to develop an AI engine, dubbed Machine Inferred Code Similarity (MISIM), that's designed to analyze software code and determine how it's similar to other code. What's most interesting is the potential for the system to learn what bits of code do, and then use that intelligence to change how software is written. Ultimately, a human could explain what it wants a software program to do, and then a machine programming (MP) system could come up with a coded app to accomplish it.