Bluehouse Group | Would you Fly with GitHub Copilot?

Four different AI generated images that were created under the prompt “If GitHub CoPilot were a person”. The first image is a realistic women smiling while wearing a sweatshirt, the second image is a cartoon of a boy with glasses smiling, the third image is a cartoon of a man with a baseball cap and a beard, and the fourth image is of a balding man with glasses and a beard.

With DALL·E 2, copy.ai, and the plethora of AI tools hitting the market, there is one tool in particular that is making us equally cautious and intrigued.

GitHub Copilot is a real-time code suggesting engine that enables code completions built directly into your editor. From short function calls to entire classes, it’s much more than just a spell-checker. The code it suggests is theoretically all available in the public domain, pulled from various open source projects that Copilot was trained on.

In theory this tool is revolutionary for our industry, but it has spurned lawsuits and copyright discussions, not to mention various ethical concerns.

Ethical issues notwithstanding, our team has mixed thoughts on this topic. On one hand, it could save vast amounts of time, on the other, it introduces a lot of code without the immediate need to understand how said code works, potentially making the it harder to maintain. Additionally, Copilot shifts the development process away from writing code, to **reviewing **code, and doesn’t necessarily guarantee a net gain in productivity.