Software Codes will be written by Machines by 2040 - According To Research

Machine Generated Code (MGC) In 2040 - Future Of Coding

Researchers at the "Oak Ridge National Laboratory" in "Tennessee" made interesting considerations about, "the machines will write most of their code by 2040". Their document requires a better understanding of the "machine's mastery" of the problem, the final piece of the puzzle.

Imagine a scenario in which a "programmer" must follow a couple of proven methods to write code which becomes a part of the larger program, which needs some sharp input from another "programmer". 

So, is the first programmer really necessary? Can we not find a "robotic" substitute for this?

> What is Automatic Programming? 

In computer science, the term "automatic programming" identifies a type of computer programming in which some "mechanisms" generate a computer program to allow human programmers to write code at a "higher level of abstraction".

> What is Code Generation? 

In computing, "code generation" is the process by which the "compiler's" code generator converts an intermediate representation of the "source code" into a form (for example, "machine code") that can be easily executed by a machine.

In the past, the "CEO of GitHub" already made a prediction, "the future of coding is not a code at all".

The document (PDF) labelled "Human also write the code in 2040 and what does it mean for the extreme heterogeneity in the computer?" 

Researchers say, the progress of programming and ongoing research efforts will result in the code generated by Machine (MGC) becoming commonplace as "artificial intelligence" in today's devices.

This is not surprising, given the recent development of programs such as:

• Microsoft's DeepCoder.
• Google's AutoML.
• DARPA Probabilistic Programming for Advanced Machine Learning (PPAML).
• AutoML.

 "DeepCoder" already uses "machine learning" to produce executable code.

With tools like "DOG4DOG", complete knowledge bases could be generated. 

In addition, technologies such as "Eclipse Modeling Framework" and "Sirius code generation" could generate the entire "data hierarchy", the "user interface" and the "intermediate layer".

It is not hidden that the "application programming interfaces" in scientific libraries are also standardized. 

So, what's missing? The document requires a better understanding of the problem domain by machines like the final piece of the "MGC puzzle".

Other important requirements for "MGC" to become a common practice would be the most efficient languages for "machine-to-machine" communications and the allocation of hardware resources to write code. 

The document also describes a number of important research directions which can be undertaken in the near future.

What are your opinions on this next big change due to the evolution of the "Machine Generated Code" (MGC)?