We’re still far from the advanced AI depicted in movies like Bicentennial Man and I, Robot. Since the launch of ChatGPT in December 2022, we’ve seen continuous advancements in its capabilities. While there are numerous articles discussing how GenAI is transforming software development, much of this information is anecdotal or theoretical.
To understand the current reality of GenAI in software development—its prevalence, applications, and effectiveness—we decided to gather direct insights from those in the field.
At NetForemost, our extensive engineering team, working across hundreds of client projects from Fortune 500 companies to emerging startups, is a valuable resource. We surveyed over 500 software engineers with diverse levels of experience to gain a clearer picture of GenAI’s impact on software development. Here’s what we discovered:
GenAI Is a Key Component in the Software Development Process
- 72% of engineers are incorporating GenAI into their software development processes.
- Notably, 48% of these engineers use GenAI daily.
- 81% utilize it to write code that they previously crafted manually.
- However, 40% of engineers feel that GenAI hasn’t significantly freed up time for other tasks.
In contrast, a recent Thomson Reuters survey across various white-collar professions found that only 12% of workers are currently using GenAI, with 11% planning to adopt it. The rest are either still considering its use or have no plans to implement it. In the software engineering field, however, adoption is much more widespread.
The rapid shift is remarkable. Just two years ago, AI was scarcely used by developers, but now it is widely adopted. Among those using GenAI, 87% engage with it daily or several times a week.
Regarding code generation, GenAI excels at creating new code snippets or “scaffolding” but is less effective at developing complete systems or integrating existing code. For example, while AI might generate a transmission component, it lacks the capability to integrate it into a functioning engine, much like building a car where the AI can provide parts but not assemble them.
GenAI Is Giving Engineers a Productivity Boost
The adoption of GenAI has noticeably accelerated product development. According to the survey, most engineers report significant productivity gains since integrating GenAI into their workflows. Specifically, 23% of GenAI users have experienced a productivity boost of 50% or more, while 71% have seen an increase of between 10% and 25%. Only 6% of engineers reported no change in productivity after adopting GenAI.
The roles benefiting the most from GenAI include Site Reliability Engineers, DevOps, GIS Developers, and Project Managers/Scrum Masters, with productivity increases ranging from 40% to over 50%. Data Scientists have reported an average productivity boost of 32%, while Full-Stack Developers have seen a 27% increase.
In terms of work quality, 74% of engineers believe GenAI has enhanced their output to some degree. 24% feel there has been no change, and just 2% think that GenAI has decreased the quality of their work. Over half (53%) of engineers reported an improvement in work quality of 10% to 25%.
Much like how Grammarly uses AI to offer writing suggestions and best practices, engineers are utilizing similar tools to enhance code quality. Although these suggestions aren’t always perfect, they frequently contribute to better code and increased efficiency.
As AI continues to evolve, software engineers are developing a collaborative relationship with it. GenAI is becoming an essential tool, much like search engines, improving efficiency while still requiring human oversight.
Software Engineers Are Turning into Editors
Just like human-written code, AI-generated code is not without its errors. According to the survey, 47% of engineers encounter minor errors in every piece of AI-generated code they review, while 16% face significant errors every time. In total, 63% of engineers find errors in every AI-generated code instance. Although we are still some distance from achieving perfection, AI allows engineers to adopt an editor’s approach to coding, rather than being bogged down by repetitive tasks.
For engineers with over 8 years of experience, 49% encounter minor errors in every piece of AI-generated code they review, compared to 39% of less experienced engineers. This suggests that, much like manual coding, the ability to identify errors in AI-generated code is closely tied to experience and seniority.
What Is AI Not Good at Doing for Software Development?
Get ready for an unexpected insight. Despite the diverse opinions in our survey, 20% of engineers believe AI struggles with code generation. So why are so many using it if they think it’s ineffective? The answer is speed. Even if GenAI doesn’t produce flawless code, its ability to quickly generate code makes it valuable. Like a chef who prefers to adjust a recipe rather than create one from scratch, engineers find GenAI’s efficiency beneficial.
What does this reveal? It seems that, more than in other fields, software engineers are leveraging GenAI for core functions despite its imperfections. GenAI is rapidly changing the nature of their work, shifting engineers from routine tasks to more complex roles.
As GenAI handles repetitive tasks, the demand for creativity, problem-solving, critical thinking, and communication skills will grow. Engineers will move from memorizing commands to collaborating on intricate challenges. The role of a software engineer is evolving from a technician to a project orchestrator.
