Access and Feeds

Coding: Using AI to Translate Between Programming Languages

By Dick Weisinger

Code migration. Translating computer code from one programming language to another is tedious, time consuming, and expensive. When migrating from legacy systems or legacy ways of operating, it often can’t be escaped. Code translators are also often used to rewrite code from a higher-level language into into a lower-level one, like the preprocessing of Typescript into Javascript.

An example of the costs of code migration is Commonwealth Bank of Australia. The bank spent $750 million over the course of five years to convert their COBOL-based system to Java.

But just like AI has been able to make translation between natural languages, like English and Chinese, AI can translate computer code too.

Recent work by a team at Facebook has created a competent transpiler, a source-to-source compiler that can translate programming languages. The initial results of the project are impressive.

TransCoder: Facebook’s Unsupervised Programming Language Translator by Rohan Jagtap

Justin Gottschlich, principal scientist and director at Intel Labs, said that “AI-based transpilers are really just trying to eliminate the minutiae of creating software. They could help accelerate productivity … [by taking care of] bugging. [And they could] increase the number of jobs [in tech] because people who don’t have a programming background will be able to take their creative intuition and capture that via machine by these intentionality interfaces.”

Digg This
Reddit This
Stumble Now!
Buzz This
Vote on DZone
Share on Facebook
Bookmark this on Delicious
Kick It on DotNetKicks.com
Shout it
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Leave a Reply

Your email address will not be published. Required fields are marked *

*

12 − 4 =