As AI's capabilities in coding continue to advance, the structure of software development teams is undergoing significant transformation, with the prospects of junior developers and Quality Assurance (QA) roles becoming increasingly precarious.
More and more Chief Information Officers (CIOs) and development team leaders are indicating that the widespread application of AI assistants will lead them to reconsider the composition of their teams. Future teams will primarily rely on AI experts and senior developers to oversee AI-generated code. Anna DeMao, a former development team leader at Fermata Energy and now a climate tech strategy consultant, noted that future application development teams will be more streamlined, with remaining senior developers focusing on how to best translate product requirements into software development. She pointed out, "When you have a large team, there are always excellent A players, average B players, and even C players, but these situations will be more pronounced in the age of AI. AI makes it more difficult to be a B or C player to some extent."
Image source note: The image was generated by AI, provided by the image authorization service provider Midjourney
In the future, the role of developers will shift to that of "editors." DeMao mentioned that some client companies have already started reorganizing their development teams around AI, with more senior developers or software architects taking charge of supervising and adjusting AI-generated code. She used the process of publishing a novel as a metaphor for this change: "Coders are no longer writers but editors. They must understand the content and who the audience is—in this case, the audience is the customer, and what our goal is."
Future development teams might consist of a product manager or business analyst, a user experience designer, and an architect using AI tools to generate prototypes, while AI would handle other software development roles, including security and compliance reviews. David Brooks, Senior Vice President at Copado, predicted, "At some point, existing software development positions will disappear, with junior software developers being the first to go." He added that software architects would reduce coding work and engage more in high-level system design and oversight of AI-generated solutions.
Although it is unclear when this transformation in team structure will reach a critical point, a recent GitHub survey shows that AI coding assistants are already quite popular among developers. Over 97% of developers from four countries reported using AI coding tools in their work. GitHub reported in January that its C 0pilot coding assistant had 1.3 million users, a 30% increase from the previous fiscal quarter. By the end of July, over 77,000 organizations had adopted C 0pilot.
Meanwhile, a Pluralsight survey showed that about two-thirds of IT professionals are concerned that AI will render their skills obsolete. Although some observers believe the impact of AI will be a long-term process, many development teams are still working to enhance their ability to leverage AI.
Ed Vatal, founder and chief consultant of IT consultancy and service provider Intellibus, said that over the next 1 to 2 years, the size of development teams might expand as more coaches are needed to enhance productivity and team AI prompt engineering skills. However, in the long run, the size of development teams might shrink as three software engineers could accomplish what used to require five or six people.
At the same time, traditional development teams will also face disruption, with more employees able to use AI and low-code/no-code tools to write applications, even if they may not fully understand how AI-generated code works. Vatal noted, "They have the ability to write code, even if they might not deeply understand how AI-generated code operates."
While many IT leaders predict that AI coding assistants will eventually lead to a reduction in developer positions, some question the rationale of delegating most programming tasks to AI. Some development leaders express concerns about AI's dual role in writing and debugging code.
There is a view that some organizations may overestimate the efficiency of AI coding assistants. Marcus Merrell, Chief Test Strategist at Sauce Labs, pointed out that a 30% increase in developer productivity is a good start but not a fundamental change. He said, "What I see is that teams think they will reap significant benefits from these tools, so they are overly aggressive in tool investment, structural and process changes, and even over-implementing planned layoffs."
Merrell believes that generative AI will not replace developers' jobs, but rather low-code/no-code tools will have a greater impact. He predicts that AI coding experiments will continue to achieve moderate success, but ultimately, large AI companies need to recoup their massive investments. He stated, "We will work hard over the next two to three years to tap into the productivity and wonders of this technology, then very slowly admit that it's actually a shell game. What worries me is that we will become dependent on these tools, and then these companies start charging the real costs required to operate these models, which will cause a huge shock to the industry."