Programming has often been considered one of the few safe jobs out there that cannot be automated away.

Now, that may no longer be the case, thanks to a new artificial intelligence designed by researchers at the Massachusetts Institute of Technology and Microsoft Corp.

Dubbed “DeepCoder,” the new AI can supposedly create new programs with requested features by simply cutting and pasting code. In a research paper, the team behind the AI noted that there have already been a few AI programmer projects out there, but they haven’t exactly been designed to be user-friendly.

“A dream of artificial intelligence is to build systems that can write computer programs,” the research team said in a paper detailing their methods. “Recently, there has been much interest in program-like neural network models, but none of these can write programs; that is, they do not generate human-readable source code.”

The researchers’ ultimate goal is for non-coders to be able to tell DeepCoder exactly the type of program they need, and then let the AI handle the rest, greatly reducing development time and cost.

“All of a sudden people could be so much more productive,”said Armando Solar-Lezama, an associate professor whose work was cited in the paper, told New Scientist. “They could build systems that it [would be] impossible to build before.”

Those systems also wouldn’t require nearly as many programmers to build. “The potential for automation that this kind of technology offers could really signify an enormous [reduction] in the amount of effort it takes to develop code,” he said.

Indeed, while a tool like DeepCoder would certainly be welcomed by pretty much any company that needs its own software, it could spell bad news for coders. Fortunately for them, DeepCoder is not able to write entirely new code on its own. Instead, it pieces together already written code taken from the source code of various existing software programs.

DeepCoder is also not creating high-end enterprise-grade software or even new mobile apps. So far, DeepCoder has only been able to write code that meets the requirements of “the simplest problems on programming competition websites.” The researchers noted in their paper, however, that they expect many extensions of DeepCoder in the future, particularly in the realm of natural language processing.

Photo: Christiaan008 Source code security plugin via photopin (license)