The research surveyed 555 software executives around the world whose companies span IT consultancy services, manufacturing, banking, financial services and insurance among others. 84 percent of respondents reported that their organizations first began to incorporate AI technologies in their SDLCs between six months to five years ago, with the earliest adopters primarily being IT services companies. Across regions, EMEA and North America remain roughly on equal footing, while APAC is steadily catching up.

The findings show that testing, quality assurance, and security vulnerability detection are by far the most widely adopted use cases for AI in software development. Nonetheless, generative AI (genAI) is set to transform the industry by significantly enhancing these processes and introducing unprecedented capabilities.

Seventy-five percent of software executives have seen up to a 50% reduction in development time by implementing AI and automation.

Early adopters are planning to increase their use of AI in other stages of the SDLC, such as user interface design, code generation, DevOps optimisation, and application maintenance. Nearly all respondents are planning to increase their investment in AI-augmented SDLC management over the next two years, indicating that AI will play a central role in driving innovation and competitive advantage in the software industry.

“AI is redefining the impossible,” said OutSystems CEO and founder Paulo Rosado. “I’m laser-focused on helping teams compress multi-year legacy modernization projects into just a few months. The latest AI disruptions have brought us the potential to compress these development timelines into even shorter and faster projects. With AI, historically impossible transformation projects are not only possible but easier, cheaper, and faster to accomplish.”

Seventy-one percent of respondents are planning to incorporate AI into application development and SDLC management workflows.

“Right now, the developer’s role is shifting from code writer to code reviewer,” said OutSystems co-founder and AI project manager Rodrigo Coutinho. “Large language models (LLMs) are a big help, but they still make mistakes. But as these models evolve, and trust in the resulting code improves, the developer’s role will be more akin to that of an orchestrator and acceptance tester of AI-generated outputs.”

Despite being a nascent technology a couple of years ago, the report found that confidence in the quality of AI-generated code has risen substantially—half of respondents said that the implementation of AI has improved software quality, enhanced decision-making, and increased efficiency in software testing and quality assurance.

But confidence is also paired with risk awareness surrounding tech debt in the form of orphan code and hallucinations, a lack of context for an organization’s specific coding needs, and scalability concerns. With strategy baked into AI in SDLC processes, 56% of respondents said they experienced or expected to experience a higher quality of applications, with fewer bugs and improved performance.

Data privacy and security concerns remain the primary barriers to broader adoption.

The AI opportunity is undeniably huge, but its wider adoption in other areas of the SDLC beyond software testing and vulnerability detection still face some barriers. Chief among these are data privacy and security concerns (56% of respondents) and regulatory and compliance challenges (42%). Moreover, 38% of executives cite difficulties integrating generative AI into existing workflows as the primary barrier to adoption.

“There’s a lot of speculation on what will change with the rise of genAI,” said KPMG US managing director Michael Harper. “While there will be challenges, those with effective change management initiatives will reskill and upskill their workforces, leading to AI and jobs evolving in tandem.”

One-third of respondents said they had a backlog of between 150 and 800 use cases for genAI.

The speed and sprawl of AI, namely genAI, is paving the way for an increase in investments for nearly all respondents.

But risks concerning the reliability of AI-generated code persist, though they can be mitigated with existing approaches, such as user acceptance testing, unit testing, and regression testing. “It’s up to the developer working with AI to guarantee the quality of the deliverables, but this becomes way more efficient with AI,” said Coutinho. “AI is, in fact, a great partner in creating tests in synthetic data.”

Other oft-cited concerns were the limited availability of skilled personnel and difficulties integrating genAI into existing tech stacks and workflows. Fears of job losses are high as well, with 89% of respondents claiming that certain roles will be eliminated by AI. This falls in line with a broader industry trend over the last couple of years. However, in the longer term, AI may well create more jobs than it displaces, resulting in a new type of developer, equipped with specialised AI skill sets.

For more insights from the 2024 survey AI in software development: Exploring opportunities and uncertainties,download the report here.