This week the arrival of Github Copilot has caught the eyes of the software developer community by its rather impressive functioning although still imperfect, but was the community concerned as it should?

OpenAI and GitHub have announced the technical preview of an AI that promises to be your new pair programmer. Copilot works as an extension to IDEs, so far compatible with VSCode, similar to IA-powered code completion tools like Tabnine, however with fundamental differences.

How does Copilot work?

The tool is the result of an investment of $1 billion from Microsoft at OpenAI, a company specialized in the development and research of Artificial General Intelligence. “Copilot automatically fills in code snippets, suggests new lines, and can even write entire functions based on the description provided”, said Github CEO, Nat Friedman, in a blog post.

Github Copilot

At its core, Copilot uses a new code generation algorithm from OpenAI called Codex. OpenAI CEO, Greg Brockman, describes it as a descendant of GPT-3, but with a focus on code generation. The Codex algorithm was trained on terabytes of public code extracted from GitHub and StackOverflow. This gives the tool the ability to write context-based code with unprecedented precision.

You can have fun with some AI suggestions for philosophical codes on Max Woolf’s blog post.

def should_terminate(Person):
    """Check whether a Person should be terminated"""

    if Person.id > 10:
        #terminate
        return True
    else:
        #continue
        return False

The Inevitability of Automation

But are the intentions of Microsoft (a multi-billion dollar company) purely benevolent and pro-community? As much as a company is community-friendly, the principle of capitalism is profit, and the profit made by eliminating the highly skilled and expensive programmers outweigh any publicized friendly benefits.

At first you can reject the idea that your job is going to be taken over by a code-generating AI, after all a programmer had to create the AI code in the first place, right?

Wrong. We tend to stick to a comfortable lie to have a false sense of security, but in reality cognitive automation has been going on for a long time, as explained in this excellent video from CGP Grey channel.

No matter how specialized a workforce, it ends up being replaced, automation is inevitable. After all, no one needs a top-of-the-line solution 100% of the time, often a mid-range solution is enough. Several similar examples have already taken place, so in an era where code can be generated by AIs, entry into the workforce will become increasingly difficult.

The dystopian truth

The parallel between the inevitability of automation and what the AI developed for Github Copilot represents is very well drawn in this video from the Fireship channel. And after analyzing the past, it becomes much harder to believe in an optimistic future.

For now, the tool will be used to help programmers perform their role better, but in the not-too-distant future AI will have learned and surpassed its own creators. Whether we are indeed “digging our own pits” toward a dystopian future for programming jobs, only time will tell.

But on the bright side we will have more free time 😄

Further reading