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.

Recent articles indicate AI is increasingly enabling non-technical individuals to code and is seen as capable of coding on its own, potentially reshaping the value of traditional STEM education. However, there is still no specific information on LLMs writing binary code directly or bypassing traditional programming languages. Instead, new concerns have emerged regarding the evaluation and potential flaws in AI-generated code, challenging the assumption of equivalent performance.