An AI-generated PHP interpreter named Phargo, written in Rust, successfully renders a WordPress front page and passes 17.4% of PHP's test suite. This marks a significant milestone for AI in creating functional programming tools, showcasing its potential in language interpretation.
Phargo is a newly developed PHP interpreter entirely constructed in Rust. Unique to this project is that it contains no actual PHP source code, relying on AI for its code generation. Despite the author's lack of experience with Rust, their AI's output rendered a basic WordPress site.
To evaluate Phargo's functionality, the experiment utilizes the standard PHP test suite comprising nearly 22,000 tests established over three decades by the PHP internals team. This approach ensures that the AI does not self-evaluate, creating a reliable measure of performance.
Currently, Phargo has achieved a pass rate of 17.4%, successfully completing 3,844 out of 22,037 tests. This score represents substantial progress from the initial performance, which began at zero. The expected maximum pass rate is around 40-45%, factoring in tests related to C extensions that fall outside the project's scope.
The Phargo project illustrates the potential of AI-assisted coding and development tools in creating functional software from scratch. While the 17% success may seem modest, it showcases a viable path for AI-generated interpreters, hinting at future improvements. The results highlight a growing intersection of AI with traditional programming paradigms.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
An AI-generated PHP interpreter named Phargo, written in Rust, successfully renders a WordPress front page and passes 17.4% of PHP's test suite. This marks a significant milestone for AI in creating functional programming tools, showcasing its potential in language interpretation.