GitHub today revealed it is adding support for both the Claude Sonnet 3.5 large language model (LLM) from Anthropic and the Gemini Pro 1.5 LLM from Google to its artificial intelligence (AI) platform for writing code.
In addition, the company has made available a technical preview of GitHub Spark, an application development tool that leverages generative AI to enable applications to be built in plain language rather than, for example, using Python or JavaScript.
Announced at the GitHub Universe 2024 conference, GitHub is now making a concerted effort to ensure application development teams that have adopted GitHub Copilot are not locked into any one specific LLM. Previously, GitHub Copilot only provided access to the LLMs provided by Open AI, which GitHub Copilot has been updated to add support for the latest o1 Preview and o1 Mini LLMs.
GitHub CEO Thomas Dohmke told conference attendees that it can no longer be denied that open-source software has won. At the same time, AI application development is evolving beyond the current AI infusion stage to an AI native era where capabilities are not added to tools and platforms but are instead deeply embedded in a way that can’t be separated.
Conversational AI tools will also be supplanted by AI agents that have been trained to automate specific tasks, he added. For example, As part of that effort, GitHub has added two AI agents for ideation and code repair to Copilot Workspace, a platform for automating the development of software engineering workflows that are still being tested. In total, GitHub has now developed five AI agents, including one that automates upgrades of Java runtimes, In addition, GitHub now provides an extension that integrates those agents with the VS Code integrated development environment (IDE).
Finally, GitHub revealed that GitHub Models, a service through which GitHub makes multiple AI models available to developers, is now generally available.
Overall, there are now more than 137,000 public generative AI projects hosted on GitHub, with Python now having surpassed JavaScript as the most widely used programming language on the platform. It’s not clear how long that might be the case once tools such as GitHub Sparks become generally available to a pool of developers that could soon number in the billions. In effect, GitHub Sparks eliminates the need to master a programming language to build an application, noted Dohmke.
AI tools are already widely employed by many DevOps teams. A Techstrong Research survey found a third (33%) are working for organizations that make use of artificial intelligence (AI) to build software, while another 42% are considering it. Only 6% said they have no plans to use AI. However, only 9% have fully integrated AI into their DevOps pipelines. Another 22% have partially achieved that goal, while 14% are doing so only for new projects. A total of 28% said they expect to integrate AI into their workflows in the next 12 months.
Mitch Ashley, vice president and practice lead of DevOps and application development for The Futurum Group, said as a widely used platform for software development and innovation, GitHub is uniquely positioned to increase uses of AI in software development. In 2025, we will also see the introduction of smaller language models that begin to specialize, solving vexing problems in software, such as recommending vulnerability remediation, applying software patches and upgrades, and more secure and higher quality code, he added.
However, the usage of AI does not automatically lead to increased productivity. In fact, a recent DORA report suggests that in some instances the rate at which software is being deployed has slowed since the rise of AI. One way or another, however, software engineering in the age of AI will never be the same again.