Yesterday, at around 5:00 pm local time, I had my first contact with a synthetic life form…

I had signed up for a Github Copilot free trial a few days prior, and was somewhat underwhelmed when nothing happened, at least at first. Maybe the machine needed some time to get warmed up.

I use an IDE (an Integrated Development Environment, for any normal people reading this: basically a graphical user interface for software development) from a company called JetBrains. The IDE itself has been evolving (slowly at first, but now much more rapidly) over the past few decades, and is really one of the most important tools in any software developer’s arsenal. There are of course free IDEs from Microsoft and the Eclipse Foundation, which today have more features than many (very expensive) IDEs of years past (lookin at you, Borland C++, circa 1996).

But seasoned (a.k.a. elder) practitioners ought to take pride in their tools, even if they come with an annual subscription. I feel a bit like Clint Eastwood’s character in Gran Turismo, when he impresses his younger apprentice not only with the beautiful muscle car that the movie is named after, but his wide array of tools, laid out in careful and painstaking order on the garage workbench.

Anyway, back to my close encounter (if this post is replete with 70s, 80s and 90s movie references… you’re welcome :) I had been working on a small Python Module for a very specialized task (in this case, communicating over a so-called “Serial” interface with a MicroController board). I had expected this task to take around 3 hours (i.e. about half a ‘man day’).

As I started adding code to the newly created “serial.py” file, the first thing I did (as a way of announcing my intention) was add a so-called ‘import’ command. This is a bit like adding a chapter heading, with a few lines of context for what is about to ensue. Almost immediately, as if by magic, the sentience started to add lines of code. What blew my mind (and still does, as I am writing this) is how it predicted almost exactly what I needed to write. In most cases, I merely had to start typing (or provide a hint or two via a so-called ‘comment’ - something akin to a paragraph heading or epilogue), and entire chunks of code would magically appear before me.

Over the next hour, I was able to achieve my objective, at an astonishing 3x efficiency improvement! Mind you, this is with zero prior training. I have spoken with colleagues, who expect their long term efficiencies to improve by at least 500%.

Needless to say, I am hooked. I now believe that Generative AI (of which github copilot is a domain specific example) is a development that lies somewhere between the invention of the world wide web (a.k.a. Web 1) in the early 1990s and the launch of the app store ecosystem with the launch of the first iPhone in 2006.

In other words, at its least impactful, this development will lead to an explosion of new use cases, business opportunities and markets for products and services (on the iphone end of the spectrum) and at its most impactful, it may lead to a multigenerational disruption of entire industries and business models (on the world wide web end of the spectrum).

What actually ends up happening is still an open question. However, one thing is clear. Whether you like it or not, this really does change everything (again) !