AI tools like GitHub Copilot are rewiring the minds of coders. could be yours next


Many people—like, say, journalists—are concerned about what generative artificial intelligence might mean for the future of their profession. It doesn't help that expert predictions on the matter offer a confusing cocktail of wide-eyed enthusiasm, sharp skepticism and dystopian despair.

However, some workers are already living one possible version of the generative AI future: computer programmers.

“Developers have come into the age of AI,” says Thomas Dohmke, CEO of GitHub. “The only question is how fast do you get on board? Or are you going to be stuck in the past on the wrong side of 'productivity polarity'?

In June 2021, GitHub launched a preview version of programming assistance called Copilot, which uses generative AI to suggest how to complete large chunks of code as a person starts typing. Copilot is now a paid tool and a huge hit. GitHub's owner, Microsoft, said in its latest quarterly earnings that there are now 1.3 million paid Copilot accounts — a 30 percent increase from the previous quarter — and noted that 50,000 different companies use the software.

Dohmke says Copilot's latest usage data shows that about half of all code produced by users is AI-generated. At the same time, they claim that there is little indication that these AI programs can work without human oversight. “After using these tools there is a clear consensus from the developer community that it needs to be a pair-programmer co-pilot,” says Dohmke.

Dohmke says the power of CoPilot lies in how it removes complexity for the programmer trying to tackle a problem. He compares this to the way modern programming languages ​​hide obscure details that earlier, lower-level languages ​​required coders to understand. Dohmke says younger programmers are especially accepting of CoPilot, and find it particularly helpful in solving newbie coding problems. (This makes sense if you consider that Copilot learned from reams of code posted online, where solutions to beginner problems are more than just examples of esoteric and rare coding craft.)

“We are seeing an evolution of software development,” says Dohmke.

None of this means that the demand for developers' labor won't be changed by AI. GitHub research in collaboration with MIT shows that Copilot allowed coders facing relatively simple tasks to complete their work an average of 55 percent faster. This increase in productivity shows that companies can do the same work with fewer programmers, but companies can use those savings to spend more on labor on other projects.

Even for non-coders, these findings—and the rapid use of Copilot—are potentially instructive. Microsoft is developing AI Copilots, as it calls them, designed to help write emails, prepare spreadsheets or analyze documents for its Office software. It even introduced a CoPilot key for the latest Windows PCs, its first major keyboard button change in decades. Competitors like Google are making similar devices. GitHub's success could help advance this effort to give everyone an AI workplace assistant.

“There's good empirical evidence and data around GitHub Copilot and productivity statistics around it,” Microsoft CEO Satya Nadella said on the company's recent earnings call. He said he expects similar benefits to be felt among other Microsoft Copilot users. Microsoft has created a site where you can try out its CoPilot for Windows. I admit that it's not clear to me how similar the tasks you want to do on Windows are to those done in GitHub Copilot, where you use code to achieve explicit objectives.

Tools like GitHub Copilot have other potential side effects besides job displacement. For example, increasing reliance on automation can lead to more errors in the code. A recent study claims to have found evidence of such a trend – although Dohmke says that while there has only been a general increase in mistakes since the introduction of Copilot, this is not direct evidence that the AI ​​helper is causing the increase in errors. Was being built. While this is true, it seems reasonable to worry that less experienced coders may miss errors when relying on AI assistance, or that autocomplete may reduce the overall quality of code.

Given the popularity of Copilot, it won't be long before we have more data on that question. Those of us who work in other jobs will soon find out if we're ready for the same productivity gains as coders and the corporate upheaval that comes with them.