The language would likely prioritize safety features, such as mandatory visibility and explicit ownership annotations, to provide clear guarantees to human reviewers. The article questions whether such a language would be practical for human use in everyday programming or if it would primarily serve as a target for AI-generated code. It considers the trade-offs necessary to achieve higher confidence in AI-generated code, suggesting that some conveniences might be sacrificed for the sake of clarity and reliability.
Key takeaways:
- Designing a programming language for AI-generated code would prioritize unambiguous clarity over convenience.
- Reducing syntactic sugar and enforcing explicit syntax could make AI output more predictable and easier to review.
- Safety features like mandatory visibility and explicit ownership annotations could be central to the language, providing instant guarantees for reviewers.
- Such a language might not be practical for human use in day-to-day work and could primarily serve as a compilation target for AI prompts.