AI in software development

A comprehensive study conducted by G+D Netcetera, Giesecke+Devrient, appliedAI, and WeAreDevelopers in cooperation with TUM reveals key insights into this transformation, offering a data-driven perspective on the integration of AI in software development processes. The study provides insight into how these tools are used today, how they influence development practices, and what opportunities and challenges they present, especially in environments where software quality, security and long-term maintainability matter most.

Imagine walking into a modern development environment and seeing almost every screen split between code and an AI assistant. That’s increasingly the reality in forward-thinking teams. ChatGPT (used by 75.5% of developers in our study), Microsoft CoPilot (38.8%), and GitHub Copilot (25.9%) have moved from experimental tools to daily companions for many developers, transforming routine tasks into opportunities for deeper problem-solving.

The adoption journey is still unfolding.

• 61.9% are taking their first exploratory steps with these technologies
• 5.8% report full integration into their development processes

Despite this early stage, the enthusiasm is unmistakable: 85.7% of developers express high interest in AI technologies, viewing these tools as partners that enhance their capabilities rather than threaten their relevance.

What happens when developers hand off routine tasks to AI? They soar. The most striking transformation occurs when teams stop using AI for mere code completion and start leveraging it strategically across their workflow. Documentation – once the dreaded final step of projects – now becomes an integrated, almost effortless process. Learning curves for new languages flatten dramatically, and standard implementations arise in minutes rather than hours.

The productivity gains are substantial: 76.3% of surveyed organizations report increased productivity through AI adoption. This efficiency boost isn't limited to coding. Nearly all respondents highlighted AI's tremendous time-saving capabilities in code documentation, an often overlooked productivity drain.

But this productivity revolution comes with an important caveat. The most successful teams aren't blindly accepting AI suggestions but maintaining thoughtful human oversight through structured review processes and established quality standards. They're creating guardrails that ensure AI serves as an amplifier of human expertise rather than a replacement for critical thinking.

Beyond faster delivery lies a more profound benefit: better, more sustainable code. While skeptics initially worried that AI might generate sloppy solutions, the reality has proven more nuanced. When implemented thoughtfully, AI tools become guardians of quality. They spot potential errors before they propagate, suggest optimizations human eyes might miss, and ensure consistency across complex codebases.

The data supports this quality enhancement:

• 66.2% of respondents believe AI improves software quality
• 55.3% report a measurable increase in overall quality through AI assistance.
• 42.4% observe enhanced sustainability of AI-generated code

The sustainability aspect is particularly promising for organizations building systems meant to last. It’s a metric that tracks how well code maintains its functionality and security over time. AI assistants excel at identifying outdated patterns, suggesting modernization approaches, and improving documentation in ways that extend software lifespan. For financial institutions, healthcare providers, and other organizations where software reliability is mission-critical, this longevity represents significant value beyond initial development speed.

The ripple effects of AI extend beyond individual productivity into the very fabric of how teams operate. Traditional hierarchies based on coding experience are evolving as AI democratizes certain aspects of development. Junior team members can contribute to complex problems earlier in their careers, while senior developers find themselves focusing more on architecture, mentoring, and strategic decisions.

This shift is quantifiable: 40.3% of respondents identified "new career paths" as an area strongly influenced by AI integration, making it one of the most significant structural changes observed. Interestingly, our research found that AI enables non-developers – including managers, product owners, and even customers – to review and discuss code more meaningfully, transforming collaboration patterns across organizational boundaries.

This shift raises fascinating questions about developer education and career progression. How do we train the next generation when the routine tasks that once served as learning grounds are increasingly automated? Forward-thinking organizations are already redesigning onboarding programs and creating new career paths that embrace AI as part of the development journey rather than treating it as an optional enhancement.

Every transformative technology brings both promise and peril, and AI in development is no exception. The most security-conscious organizations are navigating a complex landscape of data protection concerns (cited by 60.9% of respondents as their primary concern), intellectual property questions, and regulatory uncertainties (54.8% worry about unclear legal frameworks).

The technical implementation challenges are equally significant. When comparing infrastructure options, our research revealed a clear tradeoff:

This creates a technical balancing act for teams handling sensitive code, especially in regulated industries like banking and healthcare.

The emergence of "shadow AI" practices – where developers use unapproved AI tools without IT oversight – creates particular challenges for regulated industries. Yet despite these challenges, the business case for thoughtful AI integration remains compelling. 60% of organizations report a positive return on their AI investments, with only 3% experiencing negative ROI. The key difference between successful and struggling implementations often comes down to governance—having clear policies that enable innovation while maintaining appropriate guardrails.

As we look toward the horizon, the integration of AI into development practices appears set to deepen and expand. Our research indicates AI will increasingly impact all phases of the development lifecycle, with respondents expecting particularly strong influence in implementation and codification (59.1%) and testing (48.9%), while having a more moderate effect on requirements gathering and maintenance operations.

The boundary between human and machine contributions will likely become more fluid, with AI extending beyond routine coding tasks into more complex aspects of software design and architecture. This is reflected in how organizations are preparing for change. 64.8% of survey participants indicated it's necessary to develop new competencies to navigate the technological shift that AI represents.

Organizations that thrive in this new landscape will be those that view AI not as a replacement for human expertise but as a profound amplifier of it. They'll create environments where developers can leverage AI's capabilities while maintaining critical thinking, creativity, and domain knowledge that remain uniquely human strengths.

The central insight emerging from this research highlights an important truth: AI's true power lies not in replacing human developers but in augmenting their capabilities. When thoughtfully integrated into structured processes and paired with solid engineering practices, AI becomes a force multiplier for human expertise.

For organizations building mission-critical software where security, reliability, and longevity are paramount, this finding offers a clear direction. The goal isn't automation for efficiency's sake but rather using AI to reinforce high standards through intelligent augmentation. Thus freeing human creativity to focus on the most complex challenges while ensuring consistent excellence throughout the codebase.

By embracing this balanced approach, teams can harness AI's transformative potential while maintaining the human judgment and expertise that remain essential to creating truly exceptional software. Together, human creativity and AI assistance can drive progress toward a more secure and reliable digital world.

This article draws insights from the 2025 study "KI in der Softwareentwicklung" conducted by G+D Netcetera, Giesecke+Devrient, appliedAI, and WeAreDevelopers, which surveyed executives, managers, and software developers across multiple countries.