Will Generative AI Replace Junior Developers? Evolving Software Roles in the AI Era
Even your AI railways will need conductors. (Photo by Ameer Albahouth on Unsplash)
This is the third in a series of posts, entitled “AI, The Panda Strike Way,” exploring the strategic approach we take to help our clients navigate the risks and realize the massive quality payoffs of the AI era.
In Part 1, we concluded that, with the emergence of AI tools, we can translate knowledge into value. In Part 2, we discussed the importance of knowledge engineering as providing the railways for AI tools. In this installment, we’re going to talk about how current software development roles may evolve.
There’s been a lot of talk about how junior developers should consider new careers because they’ve been replaced by AI tools. The reality is that this is a dramatic oversimplification. As we discussed in Part 1, the current generation of AI tools still produces “slop” when working autonomously. They need supervision. The assumption has been that this supervision will come from senior developers. But that’s probably overkill and an inefficient use of resources.
Meet Your New Knowledge Engineers
In the context of knowledge engineering, your senior developers (and architects) are your subject matter experts. Their main job is to curate their expertise into the knowledge base artifacts that will become your AI railways. This helps ensure that your junior developers will build effectively using AI tools. When AI goes “off the rails,” they can escalate to your senior developers—now knowledge engineers—who can take the opportunity to figure out what went wrong and update the knowledge base accordingly.
Crucially, your “knowledge base” in this context includes documentation and tests for the disparate software modules that make up the systems and applications you’re building. The documentation can be incorporated into your knowledge base—providing module APIs, usage scenarios, and so on—while the tests can be run by an AI tool to eliminate possible sources of errors or even simply to better understand the module.
Good Software Process Is Good For AI, Too
The good news is that maintaining tests and documentation is part of a disciplined software process. If you have a strong process, you probably have these artifacts ready to go. If not, now is the time to level up, because the payoff for doing so has never been greater.
Combined with a general-purpose knowledge base that contains the “lore” of your organization and systems architecture, you’ll be well-positioned to take full advantage of AI tools.
Where Are The Wins?
A lot of people seem to imagine that the goal of AI is to replace your engineers. If that’s not the case—if your junior engineers are now building your applications while your senior engineers curate knowledge—how do we justify the cost of AI in the first place? First, if your senior engineers do a good job of becoming knowledge engineers, your junior engineers will become more productive. Second, your senior engineers will be increasingly free to focus on strategic technology imperatives.
Of course, you can also just gradually reduce staff, if that’s what your business demands. However, the opportunity cost of viewing AI merely as a way to reduce costs may be quite high. You may also find that people aren’t terribly eager to make themselves obsolete. But before you consider any of these questions, you need to make sure you’re actually seeing productivity gains and not just accruing technical debt at an ever-increasing rate. We’ll consider that question in our next post.