The rise of generative AI and a weak employment market has people asking some uncomfortable questions about the future for junior software developers.
Generative AI has been a disruptive force in the 20 months or so since OpenAI unleashed ChatGPT on the world, forcing employers to rethink job roles and universities about how they teach the next generation of developers their craft.
Traditionally, junior developers start their careers by working on simple, repetitive coding tasks: fixing bugs, making small feature enhancements, and maintenance work. These tasks, while often mundane, are crucial in helping new programmers build their skills and gain experience.
Now, as AI coding tools like GitHub Copilot and ChatGPT become widely available, the landscape is rapidly changing. Junior developers now find themselves in a different kind of apprenticeship – one where AI handles the tedious aspects of coding, leaving humans to focus on higher-level problem-solving and creative tasks. While eliminating toil has always been a good thing for engineers, this wave carries significant risks.
A recent blog post by ex-Amazon and Google engineer Steve Yegge speculates that we are looking at the death of the junior developer as we know it. “A lot of people picked a bad year to be a junior developer,” he wrote. “I wouldn’t want to be just getting started in the industry today.”
A growing problem
“The end of junior devs as we know it is coming for sure,” says Quinn Slack, CEO of Sourcegraph, Yegge’s current employer. “They need to adapt and change. The role of and expectations for a junior dev will look very different in a couple years.”
This could mean junior developers using AI to get up to speed with core skills faster, or using these assistants instead of human mentors, and learning more independently. “They are upleveling fast and not worrying about AI replacing them,” Slack says.
Sourcegraph, which has its own AI coding assistant called Cody, sees this as more of an opportunity than a threat. “Everyone needs to be constantly adapting,” says Slack “This was true before AI, and it’s true in a world with AI. Junior devs need to figure out what they can do that AI can’t.
Right now, that’s a lot of things, but companies look into the future when making hiring decisions and see uncertainty or lower value relative to AI. If AI can fix small bugs or add small features with little human intervention, then why pay a junior dev to do that instead?”
A perfect storm
AI isn’t the only problem here for junior developers, but it is a force multiplier for the recent wave of uncertainty impacting a sector that had become accustomed to unfettered growth.
“These last few years have been tricky for people in tech, with those hit the hardest often being the ones who are entering or are in the fledgling stages of their career,” says Jeff Watkins, chief product and technology officer at CreateFuture. He points to Robert ‘Uncle Bob’ Martin’s prophecy that the number of developers roughly doubles every five years or so, and that exponential growth eventually turns into a sigmoid, slowing down to a crawl.
“We may or may not be at that slowdown point, but the number of fresh developers being created by both universities and bootcamps was inevitably going to become unsustainable,” says Watkins. “The big question is, have we reached that saturation point?”
Whether we have or not, Watkins believes additional pressure is being placed on junior colleagues because they’ve been forced to learn how to do the job without face-to-face mentors, both because of a global pandemic and the rise of AI coding assistants.
“It’s clear that it has hurt junior developers’ careers as they’re often the first to go when a company cuts back on its tech staff, with employers preferring to keep the engine room running with mid-to-senior developers,” explains Watkins, while organizations also take their time to investigate how effectively AI can complete tasks previously reserved for junior employees.
A case for optimism
More like this
Others aren’t as pessimistic. “To some extent, this is just what happens as an industry matures,” writes Charity Majors, the cofounder and CTO at Honeycomb. “The early days of any field are something of a Wild West, where the stakes are low, regulation nonexistent, and standards nascent. If you look at the early history of other industries – medicine, cinema, radio – the similarities are striking.”
Majors believes that software engineering is different from many of the jobs that AI is set to automate – such as customer service, research analysts, or paralegals – because creating good software is a team sport, typically involving several engineers with different experience levels.
While this doesn’t mean there aren’t things to worry about, there are ways to secure the future of the industry by ensuring junior developers learn how to use generative AI as a copilot, rather than a replacement autopilot.
Taras Glek of Chatcraft, a web-based programming assistant, also sees these developments as part of the natural path of progression. Glek grew up in Ukraine, and wasn’t allowed to use a calculator. “Then I moved to Canada where we were allowed to use calculators,” he says. “The difference was that they could ask us much harder problems. The expectations from a junior developer are going to be higher now. And that’s certainly a good thing.”
A coder’s friend
Glek believes that far from sounding the death knell for junior devs, AI can help them – and free up more senior devs to do other work while large language models (LLMs) support the training of their more junior counterparts. Junior developers often struggle with syntax, or getting runtime to accept their code. LLMs “just kind of take care of the bureaucracy for you,” he says. It’s here that Glek thinks Yegge’s reasoning is flawed. “I think he’s assuming that only seniors can evaluate outputs out of the GPT,” he says.
“Reports about the death of the junior developer may be exaggerated, but the next couple of years are going to be a real challenge for that particular demographic if they aren’t keeping pace with the demands of the sector,” CreateFuture’s Watkins says. In a way, the last part of that sentence has always been the case.
