Skip to main content

Generative AI programming tools

Generative AI in java development​

  • Amazon QDeveloper
    • Catches errors
    • suggests improvement
    • handles repetitive tasks
    • real time feedback
    • helpful suggestions
    • clear explanations

Introduction to prompt engineering​

  • Prompt engineering involves crafting precise inputs to elicit the most useful and accurate outputs from AI models.
  • A poorly phrased query might generate irrelevant code suggestions, but a well-structured prompt can guide the tool to generate a complete function or even debug an issue in your code.

Prompt engineering types​

  • Instruction-based Prompts
    • direct and specific
    • tell the AI exactly what to do
    • ideal for straightforward tasks
    • By providing precise instructions, you ensure the AI produces relevant and accurate results
  • Few-Shot Prompting
    • provides examples to guide the AI's response.
    • helpful when you want the output to follow a specific pattern or structure.
  • Zero-Shot Prompting
    • involves asking the AI to perform a task without providing any examples
    • It’s most effective for straightforward tasks that don’t require additional context or explanation.
    • This approach relies on the AI’s existing knowledge of common tasks.
    • works best when the task is universally understood or when the expected output is self-explanatory.
  • Chain-of-Thought Prompting
    • guides the AI to solve problems step by step.
    • ideal for complex tasks requiring logical reasoning or multiple steps.
    • This approach encourages the AI to think through the solution, ensuring that the response includes both the reasoning and the final result.
    • useful for tasks that involve multiple logical steps or require a detailed explanation of the process.
  • Role-based Prompting
    • assigns the AI a specific persona, which can help tailor the response to match a desired tone or style
    • useful when you need explanations or answers framed in a specific way.
  • Contextual Prompts
    • includes background information or a specific scenario to tailor the AI's response.
    • useful when the response needs to be relevant to a particular use case.

Best Practices for Prompts​

  • Be clear and specific
  • Provide context
  • Use examples
  • Experiment and iterate
  • Be concise
    • Avoid unnecessary information that could confuse the AI.

Comparing Generative AI tools​

  • Amazon Q Developer: real-time coding optimization:
    • Good for Java
    • Error Detection in IDE
    • Performance Optimizations suggestion
    • Great with AWS products / services
  • ChatGPT: conversational AI designed to assist with a wide range of topics:
    • explains complex concepts
    • troubleshoots code
    • provides high-level guidance
  • GitHub Copilot: real-time code suggestions within IDEs
    • generates boilerplate code
    • automates repetitive tasks
    • speeds up prototyping
  • DeepCode by Snyk
    • AI-driven, real-time code analysis for detecting bugs, vulnerabilities, and performance issues.
  • SonarQube Community Edition
    • free, open-source platform that provides continuous code quality and security analysis.
    • identifies bugs, code smells, and security vulnerabilities.
  • Jetbrains AI Assistant
    • refactor code, receive real-time code suggestions, and interact with an AI-powered assistant