Software engineers are more valuable than capital. AI may change that

“Finding the right engineer that fits the right company in the right stage can amplify it greatly,” said Arjun Bhatnagar, co-founder and CEO of consumer privacy company Cloaked who has a lengthy background in software engineering.

Like with many roles, generative AI could shift how software engineers function or even how valuable they are to organizations. About 17% of knowledge workers report using generative AI at work to automate coding and software development tasks, according to the recent Generative AI at Work report from future-of-work software and media brand FlexOS.

Daan van Rossum, founder and CEO of FlexOS and host and writer of the “Future Work” podcast and newsletter, says the shift toward using AI technologies like ChatDev, screenshot-to-code and “GPT for coding” foreshadows a future where the line between engineers and non-technical professionals blurs.

The recent bi-annual CNBC Technology Executive Council survey found that companies across the economy are planning to accelerate spending on generative AI software like Microsoft Copilot over the next six months. A separate survey of thousands of workers across the U.S. conducted by CNBC and SurveyMonkey found that nearly three-quarters who have used AI say the technology has made them more productive — and more worried about their job security.

“Even the best engineers will be valuable until they are not,” said van Rossum. While AI is increasingly good at coding, he says problem-solving and innovating will remain essential human functions.

“I don’t think software engineering is going out of fashion anytime soon,” said Lareina Yee, senior partner at McKinsey, which is currently deploying its own large language model, Lilli, to tens of thousands of workers.

Yee, chair of the McKinsey Technology Council, recognizes that software engineering as a talent category has been in high demand over the last decade primarily because of the increase in software applications and technology enablement across industries. “With generative AI, we still probably don’t have enough software engineers, but we might be able to feel less of a shortage,” she said.

AI is particularly adept at the so-called toil tasks of software, such as code documentation review, code generation, code refactoring and modernizing legacy software languages. “You may be able to use your AI as a power tool for your software engineers,” Yee said. “They can do the things that provide the innovation, the insight, the judgment.”

McKinsey’s research reflects this. Its study on developer productivity with generative AI tells us that AI can cut time spent on simpler tasks like code documentation in half, but the time saved decreases as tasks get more complex. Complex tasks include examining code for bugs and errors, contributing organizational context and navigating tricky coding requirements.

“I think we have to put a huge caveat that this is all what the technology can do today,” said Yee, recognizing the fast pace of innovation.

Stack Overflow, a popular resource for programmers, has seen a decrease in site visits as AI applications have inserted themselves into the workflow of professionals. Some report a decrease in traffic as high as 35% in 2023, but Stack Overflow combats that metric with a lengthy explanation of cookie recategorization, saying it only lost about 5% of traffic year over year. This could be further evidence that AI is trimming the day-to-day work for software engineers.

So what are organizations going to do with the spare time their software engineers may have? Companies could address their backlog, prioritize innovation, limit the need for workforce growth as they scale or any other number of possibilities.

Yee said there’s no right answer to this. “AI is not going to draft you the answer of what you’re supposed to do. This is truly leadership experience and judgment,” she added.

Bhatnagar, however, believes ideating, innovating and coming up with new solutions is the best way to maximize that time. “You’re as good as your worst person,” he said. “If the worst person also has time to innovate, well, your entire company’s going to innovate from that point on.”

Jeff Spector, president and co-founder of Karat, says the creative aspects of development, including problem comprehension and solution design, will take precedence over low-value boilerplate code. “They’re going to focus on integrating other concerns like security or privacy or usability or performance,” Spector said. “It allows them to kind of elevate the work that they’re doing on a day-to-day basis.”

The elevation of this work could help minimize employee churn in the software engineering space by increasing job satisfaction.

In its study, McKinsey also measured the happiness of developers at work before and after using generative AI. Those who “strongly agree” to the statement “I felt happy” at work jumped from 15% before using generative AI to 50% afterwards. Those strongly agreeing to being in a flow state jumped from 25% to 44% during the same time frame.

Bhatnagar has no doubt the field of software engineering, and technical engineering as a whole, will evolve. He predicts all the different types of engineering will coalesce into two buckets: the creative problem solvers and the deep scientists. He says the employee who will last in the field amid all the innovation is “someone who is passionate about creative problem solving or can go down the track to becoming a better scientist.”