From the minds at Rice University comes a very interesting application for deep-learning AI. Bayou is an AI that, with a very small list of keywords, can produce usable blocks of code. It does this by scanning github and learning from all of the open source code on there. What does this mean for all of us computer scientists? This means that an era where AI may be able to aid us in completion of programs is on the way. For a long time this concept has been played with but never in such an efficient manner as Bayou. With other AI programmers sometimes the parameters you would have to enter to get a usable end product would surpass the program itself in length and complexity. While my title is very clickbaity I can guarantee that our jobs are safe for a very long time. This technology is only in infancy and for large programs that interact with many APIs and databases and such the need for a human to parse what needs to be written is still very essential. Bayou is more of a proof of concept that if built out even further can be very useful for speeding up the programming process.
In the post Turning deep-learning AI loose on software development, Jade Boyd makes many fair points about the ethics of Bayou. As programmers being so connected to the internet constantly has made it almost impossible to write programs from scratch. At any slight hitch we are able to search the internet and find someone else’s solution to a problem that we are facing. This point really proves that Bayou shouldn’t be seen as “cheating” rather that it is yet another tool that can give us examples of ways to solve specific problems that we maybe would have otherwise not given thought to.
I believe that applications such as Bayou are a step in the right direction. If we can get AI to write all of the simple parts of programs it would guarantee less simple errors and can also unify the format of the program. Programs can get extremely complex and Bayou wouldn’t replace us rather aid in saving us time and therefore money in the programming process.