Software development is one arena where we are already seeing significant impacts from generative AI tools. The benefits are many, and significant productivity gains are currently available to enterprises that embrace these tools. A McKinsey study (link resides outside of ibm.com) claims that software developers can complete coding tasks up to twice as fast with generative AI.

The consulting firm’s research found, unsurprisingly, that complex coding tasks were not severely affected by the use of generative AI, so worries about AI replacing developers can be safely laid to rest. However, there are “low hanging fruit” use cases where AI can drastically speed up team productivity, and improve the developer experience.

But before we get into how generative AI tools can make an impact, let’s speak more generally about improving developer productivity with methodologies, frameworks and best practices. Generative AI is just one tool in the toolbelt.

Measuring developer productivity, a subset of employee productivity, represents a multifaceted challenge. Traditional metrics, such as lines of code written or hours worked, often fall short in capturing the intricacies of complex workflows. They may not adequately reflect the quality or broader impact of a developer’s work, and a proper evaluation may require the incorporation of external factors like customer satisfaction. It is crucial to recognize that developer productivity extends beyond mere code generation; it encompasses the delivery of high-quality outputs that consistently satisfy customers, while mitigating the risk of burnout. A burned-out developer is usually an unproductive one.

DevOps Research and Assessment metrics (DORA), encompassing metrics like deployment frequency, lead time and mean time to recover, serve as yardsticks for evaluating the efficiency of software delivery. These developer productivity metrics empower engineering managers and Chief Technology Officers (CTOs) to gauge individual and team performance accurately.

Project management tools, like the widely adopted Jira, track progress, manage tasks and facilitate contribution analysis. Implementing the SPACE framework—Software engineering, Productivity, Analytics, Collaboration and Efficiency—offers a holistic approach to software development. Key Performance Indicators (KPIs), such as story points and real-time productivity tools serve as benchmarks for consistently measuring and improving software developer productivity.

Diversifying productivity measurement beyond individual performance necessitates a comprehensive understanding of team dynamics. Collaboration platforms like GitHub act as catalysts for a culture of open communication, collaborative code reviews and easily facilitated pull requests. Such platforms not only enable team members to learn from each other but also provide a collective space for skill enhancement. The strategic introduction of new features and the consistent delivery of high-quality code not only bolster the product’s competitiveness but also contribute significantly to end-user satisfaction.

DevOps emerged as a transformative methodology that seamlessly integrates development and operations practices, optimizing the efficiency of the software development lifecycle. By fostering collaboration between developers and operations teams, DevOps aims to streamline processes, minimize lead time and elevate deployment frequency. In doing so, it paves the way for an environment conducive to continuous innovation and improvement. DevOps helps to address bottlenecks and proactively manage technical debt, which allows for a work environment that keeps developers happy and chugging along.

Engineering managers can perform regular contribution analysis and use this information to integrate new tools and address employee experience concerns, shaping a conducive environment for developer productivity. The adoption of the YES (Your Engineering Success) model underscores the importance of cultivating a positive and supportive culture within the team, fostering an atmosphere that encourages innovation and creativity. This holistic approach ensures that developer productivity is measured and optimized in a manner that not only enhances individual and team performance but also nurtures the overall well-being of the development workforce.

There are a number of ways that AI can streamline development workflows. Here are some more common use cases:

Eliminating repetitive tasks

Coding often involves simple, sometimes tedious tasks, and this is where generative AI tools tend to shine. Repetitive, routine work like typing out standard functions can be expedited with auto-complete features. Tools like OpenAI’s Codex can suggest lines of code or entire functions based on natural language descriptions. Code documentation can be sped up by helping developers automatically adhere to specific documentation formats.

Natural language interfaces

Generative AI can facilitate natural language interfaces for software development tools. Developers may be able to interact with development environments, debugging and version control systems using natural language commands, making it more accessible for those without extensive programming expertise.

Code suggestion

Generative AI can also help beginners by providing context-aware suggestions, explanations and guidance as they write code. This can accelerate the learning curve for new developers and democratize access to software development.

Code improvement

Generative AI can suggest improvements to existing code by identifying redundant or inefficient portions. This can aid in maintaining code quality and performance over time. Problems that may have been tricky to pinpoint can be found and rectified much more quickly by implementing AI-suggested solutions, which can even be done automatically.

Code translation

Generative AI can also translate code from one language to another, streamlining code conversion or app modernization projects, such as updating legacy applications by transforming COBOL to Java.

Code testing

Generative AI can be employed in creating test cases automatically. It can analyze code and generate test inputs, helping to improve test coverage and identify potential issues early in the development process.

Bug detection

By analyzing large codebases, generative AI can assist software development teams in identifying and even automatically fixing bugs. This can lead to more robust and reliable software, as well as faster development cycles.

Personalized development environments

Generative AI could help create personalized development environments that adapt to individual developer preferences and coding styles. This would enhance productivity and make the coding experience more comfortable for programmers.

Enhanced Documentation

Generative AI can assist engineering teams in generating documentation by summarizing code functionalities, explaining algorithms and providing context. This can be useful for maintaining clear and up-to-date project documentation.

Generative AI in coding works by leveraging machine learning models that are trained on large datasets of code. These models, are capable of understanding the structure and syntax of programming languages.

Pre-training the model

Generative AI models are pre-trained on massive datasets containing diverse examples of code written in various programming languages. During pre-training, the model learns to predict the next word or token in a sequence of code based on the context of the preceding words. This process allows the model to capture the syntax, semantics and patterns inherent in different programming languages.

Understanding context

When presented with a coding prompt or query, the generative AI model processes the input and uses its learned knowledge to understand the context and intent. The model considers the relationships between different code elements, such as variables, functions and control structures, to generate relevant and syntactically correct code.

Code generation

Using the learned patterns and contextual understanding, the generative AI model generates code snippets as output. The generated code is based on the input prompt and follows the structure and style of the programming languages in which the model was trained.

Adapting to user feedback

Generative AI models often have mechanisms to adapt and improve based on user feedback. Developers can provide feedback on the generated code, helping the model refine its understanding and improve future outputs. This iterative feedback loop contributes to the model’s ability to generate more accurate and contextually relevant code over time.

While generative AI in coding is a powerful tool, it is not a substitute for the creativity, problem-solving and domain expertise of human developers. It serves as an augmentation tool, assisting developers in coding tasks, providing suggestions and potentially speeding up certain aspects of the development process. Developers should use generative AI responsibly, validate generated code thoroughly and complement its outputs with their own expertise and understanding.

Imagine a programmer who is tasked with implementing a complex feature for a web application project. Facing the challenge of intricate data manipulation and dynamic content rendering, she decides to integrate generative AI into her development workflow to expedite the coding process. She begins by carefully defining the requirements of the new feature, encapsulating the core logic and structure into a coding prompt. Leveraging a generative AI tool trained on a diverse dataset of web development code, she inputs her coding prompt, prompting the model to autonomously generate a preliminary code snippet aligned with the specified requirements. This generated code includes functions for data processing, event handling and dynamic content rendering.

She enters into an iterative process of refining and fine-tuning the generated code. Through this interaction, she ensures that the AI-generated code adheres to the project’s coding conventions and architectural nuances. With the generated code now meeting her satisfaction, she integrates it into the existing codebase of the web application. Despite the accelerated development process facilitated by generative AI, she recognizes the indispensable role of human validation in thorough testing to ensure the feature’s correctness, responsiveness and reliability.

The integration of generative AI into her workflow not only expedites the coding process but also empowers her to allocate more time to higher-level design aspects, user experience considerations and comprehensive testing. This use case exemplifies how generative AI serves as a valuable ally, augmenting the capabilities of developers and contributing to the overall efficiency and quality of the software development lifecycle.

IBM watsonx Code Assistant leverages generative AI to accelerate development while maintaining the principles of trust, security and compliance at its core. Developers and IT Operators can speed up application modernization efforts and generate automation to rapidly scale IT environments. watsonx Code Assistant is powered by the IBM Granite foundation models that include state-of-the-art large language models designed for code, geared to help IT teams create high-quality code using AI-generated recommendations based on natural language requests or existing source code.