GitHub has announced its Copilot platform is now available for all businesses, hopefully simplifying coding for enterprises of all sizes.
Among the various improvements over previous iterations are a significant AI boost – GitHub claims that the OpenAI model used in the software now performs better, with a 44% improvement in the coding suggestions it makes, as well as preventing suggestions that may create security vulnerabilities.
Sine June last year, GitHub claims that its code acceptance rate has increased by 8%, from 27% to 35% with the latest codex upgrade in December 2022.
AI improvements
GitHub also claims that software developers are now creating programs 55% faster with Copilot than without it. The firm hopes that with the enterprise availability of Copilot, this speed up can now benefit whole businesses rather than mere individuals.
Businesses can purchase and use GitHub Copilot straight away, assigning seats to their various employees – they don’t need to be existing GitHub users.
The AI model used comes from OpenAI, the company behind ChatGPT. Microsoft owns GitHub, and is investing big in OpenAI, so is keen to integrate its technology with just about every product it can.
The AI can make multiple coding suggestions in real time, understanding prompts from developers in either code form or in natural language, and the resultant suggestions that it makes can be easily accepted or rejected. It also learns the individual’s coding styles and preferences and adapts to them.
GitHub Copilot works with a wide range of IDE options, including Visual Studio, Neovim, VS Code and JetBrains.
As well as accelerating the coding process, GitHub makes some other impressive claims about Copilot. For instance, it claims that 46% of code across all programming languages is written with the help of Copilot, rising to 61% in the case of Java alone.
Research conducted by the company (opens in new tab) also found that 90% of developers said they completed tasks faster with the assistance, and 73% claimed it helped to conserve their mental energy. 75% also reported greater fulfilment and could attend to more gratifying work thanks to the time saved.
Despite its confidence, AI tools that help write code, including Copilot, have come under fire recently for their error rates and, more worryingly, their security flaws.
