- DeepCoder, a machine learning system created by researchers at Microsoft and the University of Cambridge, can write its own code, New Scientist reports.
- The new system could eventually allow non-coders to describe an idea for a program and let the system build it, according to Marc Brockschmidt, one of DeepCoder’s creators.
- DeepCoder pieces together code taken from existing software to make new code, much as human coders do. It also learns which pieces of code are needed to achieve different results as it works.
Some programmers may not love the idea, but there are considerable potential benefits for the enterprise. Companies could leverage a program like DeepCoder to quickly build new software or apps even if they don’t employ coders, which could also help solve some of the current tech talent gap.
If a company does employ coders, those employees could transition to take on more difficult and challenging work, allowing the AI-based system to handle the easier work.
Automating tasks like coding has the potential to enable businesses to accomplish much more, much faster. A recent McKinsey study estimates automation could raise productivity growth globally by 0.8% to 1.4% annually.