I’ve written before about how machine learning may impact you, but after a conversation with Metal Toad CTO, Tony Rost, this week the things have gotten even more interesting — machines are not only coming to automate tasks white-collar workers are doing today, they are getting ready to replace programmers as well.
Microsoft bought another company
In June of 2018, tech giant Microsoft bought a company that few outside the tech industry would recognize, but for programmers is/was a household name: GitHub. GitHub does a great job of securing code and allowing collaboration, making it famous in software development circles as the place almost everyone has chosen to store their code. Specifically, there were more than 190 million repositories (projects) on GitHub and over 40 million users in November 2019. To drive home how massive this is: there are only 27 million (estimated) software developers in the world. So it may not be an exaggeration to say that almost every programmer in the world has an account there.
3 years later…
While the acquisition of one tech company by another tech company is nothing new, just this week, Microsoft announced a new product called “GitHub Copilot” and this product is making waves around the internet. Presented as “your [Artificial Intelligence] Pair programmer”, this product actually anticipates what a programmer will do and writes the code for them. For those of you who use Gmail, this it’s a little like Smart Compose, which attempts to write your emails for you. The program gets to know you and can sometimes be a little creepy in literally writing what I would have written before I do.
The end of programming?
While an autocomplete function is not the same as operating independently, I don’t think we are more than a decade or two away from programming changing radically from what we know today. Just like self-driving cars, change may come slower than anticipated, but unlike driving, in programming, there is no real-world to contend with, and innovation in the digital space can move quite a bit faster.
For those who want a more expansive look under the hood, Jeff Delany over at Fireship, has created a more in depth (8 minute) YouTube video, which I link to below. WARNING: it gets a little nerdy, and you WILL see code.
This may be a good thing
While the end times are not upon us, it’s likely that programming will be changed radically by the year 2030. In the same way we have access to real-time grammar and spellcheckers, it’s likely that anyone — or no one, will be able to write code. And this may be a very good thing. There is a global shortage of software developers to the tune of 1.4 million unfilled Computer Science jobs in 2020. Meanwhile, the number of graduates with Computer Science degrees are only 400,000 per year.
If tools like GitHub Copilot can enable people with less formal training to write high qualify software, the future could be much brighter. It’s anyone’s guess what the years to come will look like but if history has taught us anything it will be very different from today.