As JetBrains continues to enhance its AI code completion capabilities, developers are raising important questions about privacy, integration flexibility, and the impact on coding practices. The discussion reveals a complex landscape where convenience meets security and learning considerations.
Privacy and Network Restrictions
A significant concern emerging from the developer community is the need for local AI completion options. Many companies enforce strict policies prohibiting code from leaving internal networks, making cloud-based AI solutions impractical. While JetBrains offers local Full Line Code Completion built into their IDEs, developers are advocating for more flexible integration options, such as connecting to local or same-network completion services like Ollama.
Ecosystem Competition and Plugin Integration
The developer community expresses both hope and concern regarding JetBrains' approach to AI completion plugins. While JetBrains currently offers their AI assistant as a paid plugin alongside other marketplace options, there's discussion about ensuring fair competition. Developers appreciate the current ecosystem where multiple AI plugins can coexist, such as Continue's ability to hook into three different LLMs for various completion contexts.
Code Understanding and Quality Metrics
An interesting debate has emerged regarding how to measure AI completion effectiveness. While JetBrains reports a 35% acceptance rate for their completions, developers suggest that more meaningful metrics might include:
- Code actually committed to repositories
- Success rate in completing standardized test suites
- Long-term code maintainability
Mental Mapping Challenges
A crucial concern raised by developers is the impact of AI completions on code comprehension. There's a growing recognition that rapid AI-assisted development can lead to large blocks of code that developers don't fully understand. Some developers have adopted practices like manually typing AI-suggested code to ensure better comprehension and maintainability.
Future Developments
According to JetBrains' roadmap, the 2024.3 release will bring an advanced pipeline and new model support for all languages. The company is also working on:
- Enhanced multi-line suggestions for both cloud and local completions
- Expanded language support
- Improved model quality
- Enhanced user experience
The community's response suggests that while AI code completion is valuable, the focus should be on developing tools that enhance rather than replace developer understanding and maintain security compliance within enterprise environments.