By Philip Walsh, Senior Principal Analyst, Gartner
Amid the generative AI hype, technology leaders have high expectations about the impact of AI on software development. Vendors are making bold claims about the ability of AI, particularly generative AI, to automate software development, including AI agents promoted as “the first AI software engineer” and even a “superhuman software engineer.” Such assertions have sparked speculation that AI could reduce the demand for human engineers or even replace them entirely. AI tools are already augmenting and enhancing certain aspects of the development process, and as these tools’ capabilities continue to advance, AI will significantly disrupt how developers work. However, AI tools cannot replicate the critical thinking and problem-solving skills essential for deeply understanding complex problems and producing innovative, high-quality solutions. These creative and analytical capabilities remain unique to human developers.
The influence of AI on the software engineering role will evolve over time. In the short term, the impact of AI will remain within boundaries, but it will start pushing boundaries in the medium term and ultimately break boundaries in the long term. At each stage of this disruption spectrum, AI tools will have an increasingly transformative impact on developers. Gartner predicts that by 2027, generative AI will spawn new roles in software engineering and operations, requiring 80% of the engineering workforce to upskill.
Within boundaries: AI augments existing work patterns
AI code assistants and other AI tools are beginning to change developer workflows, assisting with tasks across the software development life cycle (SDLC). These augmentations can save developers time and effort, allowing them to focus on higher-level problem-solving. However, despite the hype, the effectiveness of AI tools remains limited.
Gartner regularly hears of organisations with low developer adoption, as developers often find these tools disruptive to their established workflows, with significant concerns about the quality and functionality of AI-generated code. The effectiveness of AI will primarily be impacted by two factors in the short term: the maturity of an organisation’s engineering practices and the seniority level of the developers using the tools. Organisations with mature software engineering practices, such as established agile DevOps processes and test automation, can achieve greater productivity gains. However, even in these organisations, developers often use AI tools narrowly for generating code, which restricts their effectiveness.
To effectively use generative AI in software development, you need to know what “good” looks like to validate the AI output. But developing this level of expertise takes practice. Senior software engineers, with their deep understanding of underlying systems, are most capable of using AI tools effectively, significantly improving code quality and productivity. In contrast, junior developers, still building domain expertise, tend to over trust AI outputs, leading to security and quality issues, thus delivering limited improvements to code quality and productivity.
Pushing boundaries: AI transforms work patterns
In the medium term, AI tools will expand beyond mere augmentation and become more “agentic”, capable of breaking down complex problems into discrete, fully automated subtasks. Developers will eventually be able to offload entire tasks by directing AI agents to perform them, fundamentally transforming work patterns and increasing efficiency.
As AI agents become more capable and human software engineers become more adept at using them, most code will be AI-generated rather than human-authored. This shift will mark the emergence of AI-native software engineering. AI-native software engineering will require software engineers to adopt an “AI-first” mindset, focusing on guiding AI agents with the most relevant context and constraints for each task. To provide context for AI agents, software engineers will break down complex problems into smaller, well-defined tasks, identify relevant information and constraints, and communicate this context through carefully crafted natural-language prompts and retrieval methods. By providing this context, software engineers will play a crucial role in guiding AI agents to generate code that meets desired specifications and quality standards.
Breaking boundaries: AI creates new types of work
In the long term, AI will introduce efficiency gains and enable new capabilities, fueling an ever-growing demand for complex, innovative software. Although AI will make software engineers more productive, organisations may need more skilled engineers to meet the rapidly increasing demand for AI-empowered software. This phenomenon, known as the Jevons Paradox, suggests that as resource efficiency improves, it stimulates demand and expands resource utilisation. For instance, fuel-efficient cars led to increased energy consumption as reduced fuel costs encouraged more driving. Similarly, the increased productivity of AI-native software engineering teams will drive demand for AI-empowered applications, thereby increasing the need for software engineers.
While the emergence of AI-native software engineering will transform how software is built, the rise of AI engineering will transform what kind of software is built. Software engineers will be tasked with building AI-empowered software that can learn, adapt and make decisions based on real-time data. Building AI-empowered software will demand a new breed of software professional: the AI engineer, who possesses a unique combination of skills in software engineering, data science and AI/machine learning (ML). Organisations will race to harness the transformative potential of AI-empowered software. To deliver innovative solutions, software engineering leaders will need to invest in AI developer platforms and upskill their teams to leverage the power of AI engineering.