GitHub launched an AI chatbot for programmers at the start of the year called Copilot Chat, but this was limited to enterprise users who subscribed to Copilot for Business. Fast forward a few months later, the ChatGPT-like programming assistant became available to individual subscribers of Copilot, who pay a monthly fee of $10, but now GitHub has announced its general availability for all its users.
Copilot Chat can now be accessed directly from the sidebar in Microsoft’s Integrated Development Environments (IDEs), including Visual Studio Code and Visual Studio. This integration is part of the paid GitHub Copilot tiers but is offered free of charge to verified educators, students, and contributors to certain open-source projects.
Shuyin Zhao, VP of product management at GitHub told TechCrunch: “As home to the world’s developers, we’ve brought to market what is now the most widely adopted AI developer tool in history. And code complete was just the beginning.”
Not much has changed with Copilot Chat since its beta release, so older users should feel right at home with the latest version as well.
Copilot Chat, just like Microsoft’s Copilot, continues to be driven by GPT-4, OpenAI’s leading generative AI model, which has been specially adapted for development-related tasks. This tool allows developers to interact using natural language, enabling them to receive immediate assistance on various tasks, such as understanding complex concepts, identifying potential vulnerabilities, or crafting unit tests.
OpenAI says that GPT 4, like other generative AI models, has been trained on publicly available data, or data that is subject to copyright or restrictive licenses, though the AI startup has recently gotten into legal trouble for allegedly using The New York Times’ copyrighted articles without authority.
However, this stance hasn’t prevented programmers from initiating class-action lawsuits against GitHub, its parent company Microsoft, and OpenAI, claiming violations of open-source licensing and intellectual property rights.
Zhao has said in a statement that codebase owners will not be able to opt out of training data, but instead, suggested that the owners should keep their repositories private to prevent them from being included in future training sets.