AI Is Coming For Software Developers’ Jobs—To Make Them Better

Welcome to my LinkedIn newsletter! In each issue of Entrepreneurship and Leadership, I'll be sharing my thoughts on scaling, leading and funding high-growth startups and what the future of innovation looks like. Subscribe?here ?to stay updated.

Thanks to artificial intelligence, software developers are rapidly becoming an endangered species. At least, that’s what a steady drumbeat of grim headlines would have you believe. AI is “eating the software industry ” as companies turn to it to save money on people. ChatGPT will replace programmers within 10 years , another warns. An even bolder and scarier prediction: It will take just five years for AI to banish human coders altogether.?

To be blunt, nothing could be further from the truth. Demand for human developers is greater than ever — in fact, there’s a severe shortage of skilled people. As our appetite for software and innovation expands, this dynamic will only grow more pronounced. But there’s another critical nuance here that’s easily overlooked: Used properly, AI tools don’t replace human intelligence, they augment it. This is a truism across industries, but it’s particularly applicable in software development. From training AI to developing prompts to vetting outputs, people are irreplaceable at nearly every stage of the software lifecycle — and will continue to be so.??

In the hands of an experienced developer, AI is fast proving an invaluable digital sidekick, not a job killer. Here’s a look at why developers need support, where AI can (and can’t) help, and important pitfalls to watch for.?????

Why fears that AI will replace developers are seriously misguided

Technological change often leads to job losses, but using AI for software development isn’t like replacing a cashier with an automated checkout kiosk.?

In stark contrast to almost every other product or service, we have a limitless appetite for technology and innovation. Computer code drives that innovation. “There’s only so much food that 7 billion people can eat,” one industry expert notes .“ But it’s unclear if there’s any cap on the amount of software that humanity wants or needs.”?

A testament to that insatiable appetite is the state of software engineering at nearly every company on the planet. Software teams everywhere have backlogs that keep on piling up. Ask any business that touches software, and they’ll tell you that meeting demand for new products and features is impossible. In one large-scale study of development, security and operations (DevSecOps) teams, almost half of respondents said vulnerability backlogs were a constant problem.

Every week at my company, we agonize over prioritizing five features to build when we should be building 10. Consumer-facing firms like banks and airlines face the same struggle as their customers clamor for newer, better app features.?

The world’s 30 million software developers, meanwhile, are being pushed to the limit. Despite layoffs in the tech sector, an acute shortage of qualified developers persists. A recent survey of developers and IT professionals found that for the second straight year, recruiting people with the right skills is the biggest challenge, with more than a third of respondents putting it at the top of their list. And in a 2021 study of some 250 software developers, 83% said they were suffering from burnout. The top causes: high workload and inefficient process.

For all of these reasons, developers themselves — the people in the trenches creating the code that powers pretty much everything — aren’t sweating AI. I’ve founded three software companies that build developer tools, and I work with a global community of thousands of developers. A recent straw poll of my 60,000+ LinkedIn followers confirmed that the overwhelming majority of developers aren’t concerned about AI taking their jobs, at all. Instead they’re hungry for new and better tools.?

How AI augments the impact of human developers?

Developers are constantly looking for ways to automate things. They see AI largely as a technology that can make their job better, not steal it, by handling the grunt work that has long plagued their industry. Importantly, however, they’re embracing these tools as digital assistants that augment human brain power rather than replace it.?

For starters, all AI needs training from a human mentor. As a result, it’s only as good as the person in charge. Garbage in, garbage out, as the coding maxim goes. Anecdotally, many of us have seen this with ChatGPT and consumer AI applications: If you don’t give it the right prompts, the answers it provides aren’t useful.?

Understanding and validating outputs also demands human expertise. Generative AI is notorious for hallucinating — in other words, making things up. If a lawyer asks a large language model like Google’s Bard to create a legal brief, the LLM may well concoct its own cases and facts. The same kind of fabrication can happen when a developer enlists AI to generate a policy for cloud governance. Human creativity and judgment are required to flag these oversights and help it improve.?

Finally, there’s an important financial calculus all companies have to make. Remember that AI is expensive , especially when striving for 100% accuracy without human oversight. In most cases, the hassle and the cost of running truly independent AI simply aren’t worth it. There isn’t much point in shelling out $5,000 to have AI do a task with 100% accuracy, when you can spend $5 for 95% accuracy and have a human take care of the rest.?

All of that said, I see AI transforming how developers work every day in profound ways. Writing code is just the tip of the iceberg. As a debugger, AI can quickly sift through millions of lines to spot errors before they become a problem, recommending fixes based on what worked in previous software builds. AI can pinpoint security vulnerabilities by checking the code against all known weaknesses, then generate fixes for them, too. To developers’ relief, it can also reduce the time to test a software update from hours to minutes, thanks to its ability to zero in on the right sets of tests for the job.

Meanwhile, one of the most useful features of AI is actually hidden in plain sight: natural language interfaces . The ability to query in “people language” rather than computer language expedites everything from designing stress tests to creating policies for cloud costs, security and risk management.

Ultimately, AI promises to put a serious dent in the innovation backlog at companies by making software teams quantifiably more productive. While these are early days, I’m already seeing a significant improvement in efficiency. That doesn’t mean people will lose their jobs — it simply means more stuff will get built. The outcome: happier customers and shareholders.?

But AI only works properly in collaboration with a person. Confession: I like to play the sci-fi video game Halo. As the protagonist, Master Chief, I have an AI assistant named Cortana. She helps me get things done and deal with the bad guys, but there’s never any doubt who’s boss. Developers — and all the doomsayers — should look at AI the same way.

Thank you for reading! I'd like to know your thoughts in the comments below. For more insights from my experience as a serial entrepreneur and how we can harness the power of software to change the world, be sure to?subscribe ?to Entrepreneurship and Leadership.