The Rise of ‘Vibe Coding’ and AI Copilots

A new approach to software creation, often dubbed “vibe coding,” is gaining traction. This method leverages AI copilots – tools like GitHub Copilot – to translate natural language instructions into functional code. Instead of meticulously writing each line, developers (and even non-developers) can articulate their desired outcome in plain English, and the AI generates the underlying code. This dramatically reduces the barrier to entry for application development.

John Moehrke of Moehrke Research LLC has been at the forefront of this movement, successfully utilizing AI to create two distinct applications. His experience highlights the potential for rapid prototyping and development, even for complex tasks. “I wrote in a readme what I wanted done, and Copilot produced a working application on the first try,” Moehrke explains. “I then spent two days improving them, adding features and robustness, without ever needing to directly touch the code.”

Two Pioneering Applications: SAMHSA ValueSet Viewer and Security Labeling Service

Moehrke’s initial project addressed a specific need within the healthcare interoperability space. He developed a SAMHSA ValueSet Viewer, a web application designed to display the contents of ValueSets from the Substance Abuse and Mental Health Services Administration (SAMHSA). This was born out of a limitation in existing tools, which struggled to render large ValueSets. The application utilizes the FHIR (Fast Healthcare Interoperability Resources) standard and the $expand operation to retrieve and present the data.

The second, more ambitious project is a Security Labeling Service (SLS) Reference Implementation. This application, built with similar AI assistance, is a Docker-deployable server with a FHIR $operation compliant API. It supports ValueSets with multiple topic values, a crucial feature for managing sensitive data. The SLS is designed to categorize and label data based on its sensitivity, ensuring appropriate access controls and compliance with regulations.

The development of the SLS involved a more iterative process, with Moehrke refining the AI-generated code and adding features based on his expertise. He leveraged existing work from the SHIFT-Task-Force and the LEAP SLS Reference Implementation, building upon established foundations.

The Importance of Provenance and Transparency

Moehrke emphasizes the importance of transparency in AI-assisted development. Both applications prominently display their provenance, clearly stating that they were developed by GitHub Copilot (Claude Sonnet 4.5) under his direction. This acknowledgment is crucial for building trust and accountability in a rapidly evolving technological landscape.

This application was developed by GitHub Copilot (Claude Sonnet 4.5) …, at the direction of John Moehrke of Moehrke Research LLC.

What are the long-term implications of this shift towards AI-assisted development? Will it democratize software creation, empowering individuals without traditional coding skills? Or will it exacerbate existing inequalities, concentrating power in the hands of those who control the AI tools?

The development process wasn’t without its challenges. Initial bugs were identified and swiftly corrected with the assistance of the AI, demonstrating the potential for rapid iteration and improvement. Moehrke’s experience underscores the collaborative nature of this new development paradigm – a partnership between human expertise and artificial intelligence.

Frequently Asked Questions

• What is “vibe coding” and how does it differ from traditional software development?

  “Vibe coding” refers to the practice of using AI copilots to translate natural language instructions into functional code, reducing the need for extensive manual coding. It differs from traditional development by shifting the focus from writing code to defining the desired outcome.

• What are ValueSets and why are they important in healthcare?

  ValueSets are collections of codes that represent a specific clinical concept, such as a disease or a procedure. They are crucial for standardizing healthcare data and enabling interoperability between different systems.

• How does GitHub Copilot contribute to the development process?

  GitHub Copilot acts as an AI pair programmer, suggesting code snippets, completing lines of code, and even generating entire functions based on natural language prompts.

• What is the role of FHIR in these applications?

  FHIR (Fast Healthcare Interoperability Resources) is a modern standard for exchanging healthcare information electronically. These applications leverage FHIR to access and process healthcare data.

• Is AI-assisted development a threat to traditional software developers?

  While AI-assisted development may automate some aspects of coding, it is more likely to augment the skills of developers, allowing them to focus on higher-level tasks such as design and problem-solving. It’s a shift in skillset, not necessarily a replacement.