OpenAI has announced improvements to ChatGPT's capabilities across low-resource and legacy programming languages, including COBOL, SQLite, and other languages with smaller training datasets. The company has created new evaluation benchmarks specifically designed to measure performance on software engineering tasks within these language ecosystems.

This development addresses a real business need: many enterprises still maintain legacy systems built in older languages, and developers often lack modern AI assistance for maintaining or migrating this code.

What This Means for Your Business

If your organization manages legacy systems in COBOL, older SQL dialects, or other less common languages, ChatGPT can now provide better code assistance for maintenance and modernization tasks. This could accelerate your ability to document, refactor, or migrate legacy systems without needing to hire specialists in outdated languages—potentially reducing costs and time-to-completion for technical debt projects.