Many people – like journalists, for example – are rightly concerned about what generative artificial intelligence could mean for the future of their profession. It doesn't help that the prognostications of subject matter experts offer a confusing cocktail of enthusiasm, trenchant skepticism, and dystopian despair.
Some workers are already living in a potential version of the future of generative AI: computer programmers.
“Developers have arrived in the age of AI,” says Thomas Dohmke, CEO of GitHub. “The only question is: how fast do you ship? Or will you stay stuck in the past, on the wrong side of the “productivity polarity”?
In June 2021, GitHub launched a preview version of a programming aid called Copilot, which uses generative AI to suggest how to complete large chunks of code as soon as a person starts typing. Copilot is now a paid tool and a resounding success. GitHub owner Microsoft said in its latest quarterly results that there are now 1.3 million paid Copilot accounts, a 30% increase from the previous quarter, and noted that 50,000 different companies use the software.
Dohmke says the latest Copilot usage data shows that almost half of all code produced by users is generated by AI. At the same time, he says there is no evidence that these AI programs can operate without human oversight. “There is a clear consensus within the developer community, having used these tools, on the need to be a co-pilot paired with a programmer,” says Dohmke.
The power of Copilot is how it eliminates complexity for a programmer trying to solve a problem, Dohmke says. He likens this to the way modern programming languages hide tricky details that earlier, lower-level languages required coders to resolve. Dohmke adds that younger programmers are particularly accepting of Copilot and that it seems particularly useful for solving beginners' coding problems. (This makes sense considering that Copilot learned from tons of code posted online, where solutions to beginners' problems outnumber examples of abstruse and rarefied coding techniques.)
“We are seeing the evolution of software development,” says Dohmke.
None of this means that developer labor demand won't be changed by AI. GitHub research in collaboration with MIT shows that Copilot helped coders facing relatively simple tasks complete their work, on average, 55% faster. This increase in productivity suggests that companies could do the same work with fewer programmers, but could use those savings to spend more on labor on other projects.
Even for non-coders, these results – and the rapid adoption of Copilot – are potentially instructive. Microsoft is developing AI Copilots, as it calls them, designed to help write emails, create spreadsheets or analyze documents for its Office software. He even introduced a Co-pilot key to the latest Windows PCs, its first major keyboard button change in decades. Competitors like Google create similar tools. The success of GitHub could help give everyone an AI work assistant.
“There is good evidence and empirical data around GitHub Copilot and the productivity statistics surrounding it,” said Satya Nadella, CEO of Microsoft. said during the company's latest earnings conference call. He added that he expects similar gains to be felt among users of Microsoft's other Copilots. Microsoft has created a site where you can try its Copilot for Windows. I confess that I don't understand how similar the tasks you want to do in Windows are to those you do in GitHub Copilot, where you use code to achieve clear goals.
There are other potential side effects of tools like GitHub Copilot besides job displacement. For example, increased reliance on automation could lead to more errors in the code. A recent study claimed to have found evidence of such a trend, although Dohmke claims he only reported a general increase in errors since the introduction of Copilot, and not direct evidence that the AI assistant was causing an increase errors. While this is true, it seems fair to worry that less experienced coders might miss errors when relying on AI help, or that the overall quality of the code might decline thanks to semi-typing. automatic.
Given the popularity of Copilot, it won't be long before we have more data on this question. Those of us in other jobs may soon discover whether we enjoy the same productivity gains as coders — and the corporate upheaval that comes with them.