This post was originally published on this site.
Steve Rodda is the CEO of Ambassador, a cloud-native development company helping enterprises manage microservices on Kubernetes.
AI can write so much code in so little time, it must be making us more productive, right? Maybe not if it isn’t writing the code we need.
The current generation of AI works by amalgamating existing data, analyzing common patterns and predicting a best-fit solution. That may be a good approach at first—but look deeper, and you’ll find it’s at odds with what makes the most impactful code and valuable applications.
AI isn’t producing elegant solutions or novel ideas because it can’t.
If you want to build software that does more of what’s already out there, then AI might be able to replace your developers. But if you want to create something unique and valuable, you must invest in your people.
Your best developers have the expertise and wisdom that AI can’t reproduce. Finding the right model for them to work with AI won’t just get you more code, it can get you something better.
What AI Can Do For Your Business—And What It Can’t
My intention isn’t to discredit the AI hype. It’s a remarkable, powerful technology, and your developers should absolutely use it. However, you need to have clear boundaries set about its limitations.
Where AI Can Help You Grow
AI can offer huge efficiency increases in the right spots—boilerplate code is a classic example. AI-assisted code generation can replace rote manual work like setting up authentication, connecting APIs, configuring providers, and creating similar plug-and-play pieces of your applications. It can do so because it’s mostly reproducing work done elsewhere thousands of times.
That’s a good thing—people generally don’t like grunt work, so automating it is a win for efficiency and developer experience. And, assuming you choose good tools, AI can do it with fewer errors and automatically apply governance rules, saving you time in debugging and QA.
This might fuel the anxiety about junior-level developer jobs disappearing—and the fear is well-founded. These jobs will likely bear the brunt of AI’s accelerating progress. What to do about this is an important question that the industry needs to tackle and has yet to remain solved. Senior developers, on the other hand, will be even more prominent players in the foreseeable future, and here’s why.
AI Can’t Create What’s Not Already There
A successful business model almost always depends on a complex, creative insight. It’s pretty rare for a company to chart significant growth through marginal efficiency gains. To hit growth goals, you need to create something new.
AI can’t do that for you, at least not yet. Sure, you can ask AI to Frankenstein together some new ideas for brainstorming purposes, but if you’re looking to transform a business, AI falls short of the creative task in several key ways:
AI can’t analyze context. You can plug in a lot of background yourself, but AI is ultimately limited in understanding.
• The cultural and social factors needed to tailor your product to different markets
• Industry-specific information that shapes the way you use data over time
• The political context influencing the legal and regulatory climate, which in turn impacts your products or practices
• Your app’s plans for growth
AI can’t think symbolically or in analogies. This includes explaining new ideas, making novel comparisons or anticipating innovative use cases.
AI can’t provide complex risk analyses. It doesn’t understand your unique users and can’t anticipate how your app should address your unique risks.
Your best developers and security engineers are doing all these things. They collaborate with your legal staff to manage risk, work with product teams to create resilient applications that can grow and change and their knowledge helps your marketing teams deliver better product education.
AI can be your code monkey if you only want basic, rehashed ideas. But for solutions that add value and stand up to headwinds, you need actual human developers. You can’t rely on AI to do the technical work that matters for the long haul.
Is This A Change For Good?
Here’s another way to consider the evolving landscape: The way to get positive results from AI is to let it do its best work so that people can do their best work.
The work of software development is changing. Rather than panic, consider that software developers are pretty unhappy with the status quo. They’re frustrated with cumbersome processes, clunky tools and outdated codebases, and they’re spending too much time waiting for cloud infrastructure to run code. What if AI could make that situation better?
Lately, much attention has been paid to the idea that developers need to spend more time writing code—that’s the crux of the controversial 2023 McKinsey productivity study. But didn’t we just decide that the code-crunching work of the inner dev loop is ripe for AI automation?
There’s no contradiction. Optimizing the inner dev loop isn’t just about spending more time writing code—it’s about finding more space for your best developers to think so that the code is quality. When tools cut down on drudgery and frustration, fewer things get in the way of experimentation, and that’s where the creative leaps happen.
Let The AI Revolution Be A Catalyst For Positive Change
In some ways, AI isn’t that different from any other productivity tool. It can help you reach your goal faster. But first, you have to know what the right goal is and be sure you’re working towards it. AI can’t do that for you—you need to collaborate with smart, creative people who know your business.
The AI boom is an opportunity to refocus your company on the innovations where you can deliver the most value. From there, choose tools that take on the most rote, repetitive parts of people’s jobs so developers can focus on more profound problems. The key to growth will be giving your devs space to use their unique talents and insights to do the work that only they can.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?