We here a lot about generative AI and how transformative it is, but how exactly can we use it to our benefit, increase our own productivity and empower ourselves. In this blog post, we look at generative AI in Coding and in which areas it will help us.
- Explaining Code and Natural Language Processing
- Code Assist and Code Completion
- Code Conversion / Refactoring
- Debugging Errors
- Democratization of Development
- Code Review and Pair Programming
- Proper License Attribution
- Documentation
- Log Summarization
- Unit Test Creation
- Security and Vulnerability Scanning
Explaining Code and NLP
Most AI-powered code assistants have an “Explain Code” feature, which is usually a matter of selecting the code, error, or log-entry and pressing a button. They then go on to explain what you selected and do so in “human” language. If you have more questions, you can follow-up using more natural language and you will receive replies in the same language. This ability to understand natural language is called NLP (Natural Language Processing).
AI-powered code assistants support most environments, and most programming languages. It doesn’t matter what your skill level is and has unlimited patience. It is available 24/7 and there are no stupid questions you can ask.
Even if you’re a senior developer, at times, you can stare blind on a relatively small issue. Something an AI-powered code assistant can pick up quickly.
Code Assist and Code Completion
AI-powered code assistants support a large variety of programming languages, are kept up-to-date and make development tasks more manageable and accessible for developers of different skill levels.
These assistants usually come in the form of IDE extensions. Extensions for Visual Studio, Visual Studio Code, JetBrains Suite of IDEs and others enable relevant and real-time context aware code suggestions, significantly reducing errors, increasing productivity and accelerating productivity.
This is particularly useful for rapid prototyping, boilerplate code, well-defined code or API integrations. It’s also useful if you need to get started, but not sure where to start. Notable AI-powered code assistants are:
- Microsoft’s GitHub CoPilot
- Google Duet AI for Developers
- JetBrains AI Assistant
Vendor specific assistants are usually pre-trained and fine-tuned for their environments. For example Google Cloud (and related services / libraries) for Google Duet AI for Developers en Microsoft Azure for GitHub CoPilot.
I personally use both through their respective Visual Studio Code extensions.
Code Conversion / Refactoring
Generative AI demonstrates remarkable capabilities in refactoring code or even seamlessly translating code from one programming language to another. This functionality is particularly valuable when original developers are no longer available, and you’re dealing with the following situations.
- Migrating existing projects
- Integrating diverse code-bases
- Modernizing code-bases
- Switching APIs
- Reducing reliance on, and moving away from legacy code
- Reducing dependencies
Debugging Errors
Generative AI can be used to find and debug errors through code analysis, pattern recognition, and contextual understanding. It can even detect errors that are not immediately apparent and provide recommendations on how to fix or improve the code. Much of this can be done in-line in the IDE through extensions and without having to execute and run the code.
Democratization of Development
The abilities of generative AI and leveraging of Graphical User Interfaces (GUIs) with pre-built components fuels the emergence of low-code and no-code platforms and empowers non-programmers to actively contribute to the creation of applications.
This trend, not only accelerates the application development process but also broadens participation in software creation, marking a significant shift in the traditional landscape of programming. Notable AI-driven low-code/no-code frameworks are:
- Buble
- Glide
- Softr
- Webflow
- NoCode Logic
- Zapier
- Integromat
- Mendix
Code Review and Pair Programming
Similarly to Code Assist and Code Completion, Generative AI helps software engineers significantly improve their code review and pair programming workflows. By applying machine learning to large datasets of open source code repositories, generative models can identify patterns and best practices for code structure, modularity, error handling, security considerations, and other aspects of software design that experienced engineers focus on during code reviews.
Through configuration or prompts, developers can specify which convention or programming style to use and AI code assistants can review code in real-time and offer suggestions in-line in the IDE through the use of IDE extensions.
Proper License Attribution
Generative AI can assist developers in automatically generating and embedding accurate attribution notices throughout the code. This doesn’t only streamline the attribution process, it also standardizes it and reduces the chances of accidental omissions while ensuring that credit is appropriately given to individuals or teams involved in the code’s creation, fostering transparency and adherence to open-source principles.
Documentation
Generative AI can autonomously produce detailed documentation, including comments, function explanations, and usage guidelines. This can be done either in-line in an IDE, for an entire file, or for an entire code-base at once. Popular
Documenting ones code greatly improves readability, future usability and cooperation with other people.
Log Summarization
Log summarization, a critical aspect of system monitoring and troubleshooting, is significantly streamlined with the assistance of generative AI. Both for error debugging and security analysis.
In the realm of large-scale data logs, which can be overwhelming and time-consuming for human analysis, generative AI models excel in distilling key insights. I use this regularly and found it useful in the following cases.
- Understanding stack traces
- Understanding nested logs
- Summarizing logs with repetitive messages
- Sifting through large (web) server logs
- Analyzing container logs or serverless function logs
These models leverage advanced natural language processing and machine learning techniques to comprehend the context and patterns within logs, automatically generating concise and informative summaries. By discerning important events, anomalies, or trends in logs, generative AI not only accelerates the identification of issues but also aids in root cause analysis.
This automated summarization process enhances the efficiency of IT professionals, allowing them to swiftly grasp the essential details from extensive log data. The result is a more proactive and effective approach to system monitoring and troubleshooting, ultimately contributing to improved system reliability and performance.
Unit Test Creation
Generative AI transforms the process of unit test creation by automatically generating test- cases.
Generative AI analyzes the codebase, identifies potential scenarios, and dynamically creates unit tests that cover various pathways and conditions within the code. By understanding the logic and structure of the program, these AI models generate test inputs and expected outputs, providing comprehensive test coverage.
Security and Vulnerability Scanning
Generative AI can autonomously analyze codebases to identify potential security risks, vulnerabilities, and areas prone to exploitation. I’m using this myself in VSC and though it’s a little resource intensive on slower machines, it works really well.
- Snyk – free for indivuals