“Artificial intelligence and software development are now absolutely linked for the rest of our lives,” commented the CEO of GitHub, Thomas Dohmke said today during a presentation at the Collision Conference in Toronto, Canada. “In a world devoured by software, every developer needs a co-pilot.”
In an interview after his talk, Dohmke expanded on this topic a bit when asked if he thinks all developers will be using AI in the near future. "I think the obvious answer to that is that the FOMO (Fear of Missing Out) in companies is already so great that they are looking at the competition and wondering if their competitor has already adapted, [GitHub] Copilot, and that means that that competitor has it integrated at 20%, 30% or 40%, the competitor has an advantage.”
Furthermore there really is no disadvantage for developers to use a tool like Copilot. "It is natural. No reason not to use Copilot,” she said. “I think it's becoming part of the standard toolset that all developers will use. Ultimately, developers who don't use it will exist, just as developers of Cobalt".
He also noted that tools like Copilot will be integrated throughout the entire development lifecycle.
GitHub's Copilot was among the first AI-based development completion services and remains the most popular, even as AWS CodeWhisperer and, more recently, Google's Bard-based ones are also being adopted by developers. As part of Dohmke's talk, GitHub also announced today some of its latest findings about how Copilot is being used by users.
One number that hasn't changed is that GitHub still says that, based on its analysis of a sample of nearly a million users, developers initially accept just under 30% of code suggestions, and the more they use it, the greater their acceptance rate. About 35% of the suggestions are achieved after six months of use. Those numbers, Dohmke believes, won't change so drastically in the near future, although he noted that 50% "would make us happy."
Image credits: GoodHub
What is perhaps just as important is that Copilot is especially useful for less experienced developers (which GitHub defined by the average number of commit actions to GitHub before using Copilot).
“As developers continue to become fluent in prompting and interacting with AI and new models that enable natural language to drive the development lifecycle, we anticipate that 80% or more of code will be written with AI, which will help to democratize software development for more people. A report from GitHub co-written by Dohmke and Marco Iansiti and Greg Richards from Keystone.ai Explain these points.
Given the explosive growth of AI and the shortage of developers in this space, GitHub also notes that generative AI tools hold great promise for making those developers more productive. The company expects that, globally, generative AI-powered development tools will add $1.5 trillion to global GDP by 2030 and that each skilled developer not incorporated into this model represents $100,000 in lost GDP. Meanwhile, generative AI development tools can compensate for about 15 million additional developers, GitHub believes (hence the total impact of $1,5 trillion). The company believes this is a conservative estimate.
