I have a few friends who think that AI will be the end of their software engineering career.
Honestly, I don’t agree with them.
I started working in the 90s and we too had a secret which powered our careers, made us 10x engineers and allowed us to stand out from the rest.
And no, this wasn’t even Google. Before Google, and before the internet there was Usenet.
Usenet was a series of newsgroups – mailing lists essentially. You could subscribe to a newsgroup (via an NNTP server) and then have all messages on those groups forwarded to your email address. So essentially it was a mailing list where you could ask questions and get answers. Usenet predated the modern internet but it became super-powered when search engines came along.
Suddenly archives of Usenet groups became searchable, browsable. Suddenly services like DejaNews appeared. Altavista and then Google made access simpler.
Didn’t know the answer to something? Google it. Before then, DejaNews it. These were the tricks that allowed me to get ahead, land the consulting gigs, become a better coder and a better engineer. I learnt more. I was more confident.
So if you see AI as a threat to your engineering career, ask yourself, what would you rather do? Laboriously write out all the code by hand like we did back in the olden days? Or be able to concentrate on actually solving business problems by asking other experts what you should do?
AI coding assistants are just that. Assistants. There is no wholescale creation or support of systems by AI.
Users are no less hard to understand and provide for than they were in the 1990s. Problems are not simpler to fix than they were in the 1990s. While observability helps we’re still as an industry generally woeful at operations and telemetry. Problems with scaling are still hard. Edge cases are still plentiful. Software is now also on another scale. Honestly we need all the help we can get, and AI currently isn’t even good enough to be a consistent help.
With the sheer amount of software in the world, and systems that rely on it, there’s plenty of room for another person to help keep them working.
So don’t fear AI. Embrace it. Enjoy the opportunities it affords us to do less menial work and more of the fun thinking work. The stories which say the opposite, and they are legion, are focussing on the nuts and bolts of programming. Yes, AI might be able to write a class, a test or even a simple program for you. It might be able to spot and improve a subtle bug. Can it solve operational problems as well as humans? Even if it can, what would that even look like and can it predict where and how the next failure will occur?
Working with AI models either integrated like GitHub Copilot or in a more standalone way with local or hosted models will only enhance your career. Every career benefits from learning to work with the emerging technology and not against it.