Large language models (LLMs) will be capable of writing binary code directly, bypassing traditional programming languages, with performance at least equivalent to code written using programming languages, by May 8, 2028.

Resolved positively if a peer-reviewed academic paper or a widely recognized industry benchmark (e.g., from Google, Microsoft, OpenAI) demonstrates an LLM generating binary code directly that achieves performance metrics (e.g., speed, memory usage, functionality) demonstrably equivalent to or better than human-written code in a high-level language for a non-trivial task, as verified by independent experts, by the resolution date. The LLM must not use an intermediate high-level programming language as part of its generation process.