GitHub Copilot: Accelerating AI’s ‘Time to Value’ in Programming

Maria Diaz of ZDNET dives into the topic of generative AI’s ability to help coders in a featured article. David Gewirtz of ZDNET has found from his first-hand experiments that OpenAI’s ChatGPT “can write pretty good code.” On the other hand, some studies have found large language models such as GPT-4 are well below the overall level of code quality compared to human coders.
Inbal Shani, the chief product officer for GitHub, mentioned in an interview with ZDNET that the real change of the world of generative AI is the creation of a new abstraction layer on top of AI, that being natural language. The abstraction layer has initially been used for code completion but has the potential to broaden out to many other uses of AI beyond code completion.
GitHub introduced GitHub Copilot, its version of code assistance, in June 2021. It’s been a transformational year for AI in programming. GitHub has over a million paying customers using Copilot and over 37,000 organizations using it, says Shani. Prominent users like Accenture have seen that Copilot helps reduce repetitive “boilerplate code.”
Accenture has retained 88.5% of the code written by Copilot. It has led to increased productivity in the completion of pull requests and builds. “Sometimes, developers hold themselves back from doing builds, but using Copilot has helped build that trust,” Shani noted.
Automating tasks with AI gives developers the opportunity to invest time in other areas. It can free up time by reducing the need to switch between different tools, which contributes to the overall happiness of developers. Additionally, the tool is finding applications in hiring, onboarding, and peer programming situations.
Microsoft is working with customers such as Accenture to improve Copilot’s utility. They aim to create a personalized experience for developers and are preparing to release an enterprise version of Copilot in February. This version will cater to enterprise needs and will include features such as personalized AI models and additional testing capabilities. AMD is one of the beta customers for the enterprise edition, aiming to fine-tune internal generative AI models.
Overall, AI in programming seems to be having a positive impact on developer productivity and happiness, but the process is ongoing, and there is still much more research and development to be done.