As reported by New Scientist, DeepCoder can search for the best source code combinations and integrate them to solve issues. By using program synthesis, the AI can separate good code and combine it to create new applications and solve challenges. DeepCoder uses information from a human programmer to get the best results. It can search quickly after receiving information such as available inputs and what the programmer wants the result to be. A new application can be created much faster than the capabilities of a human programmer. As it uses machine learning capabilities from Microsoft Research, the system learns as it searches. This allows it to be the fastest such system so far. Moreover, it will become faster the more it learns and remembers. We are used to the idea of AI taking over aspects of our lives and automating them. Microsoft says DeepCoder could take over the mundane programming tasks and complete work more quickly. Human programmers will then be able to undertake more work. Speaking to New Scientist, unrelated researcher Armando Solar-Lezama says productivity will increase through projects like DeepCoder: “All of a sudden people could be so much more productive. They could build systems that it [would be] impossible to build before. The potential for automation that this kind of technology offers could really signify an enormous [reduction] in the amount of effort it takes to produce code.” The full research paper for DeepCoder is available in PDF.
Beyond DeepCoder
The advancement of artificial intelligence has caused controversy. Many are worried AI will become to self-sufficient and will replace jobs, leading to an employment crisis. Whether these worries are justified will only be decided by time. Microsoft has been a leading name in AI development. At the start of last year, the company became one of the first major tech companies to open source its AI software. The deep learning AI toolkit is available on GitHub.
