For decades, we've treated AI as a sophisticated autocomplete—a tool that helps us write Python or JavaScript faster. But there's a growing tension in the industry: when an LLM generates 500 lines of code in a blink, the human review process becomes a bottleneck. We are essentially trying to audit machine-speed output using human-speed reading tools.
The Shift to 'AI-Native' Syntax
Enter Jacquard. Unlike traditional languages designed for manual keystrokes, Jacquard is a programming language built specifically for a regime where machines do the writing and humans do the reviewing. Developed as a FriendMachine research project, its very structure was born from AI analysis of Abstract Syntax Trees (ASTs) from mainstream and esoteric languages.
By optimizing for the review process rather than the typing process, Jacquard aims to bridge the gap between the high-level intent of a prompt and the low-level reality of the execution. It’s not just about efficiency; it’s about trust and verifiability in a world of 'AI slop.'
A New Era of Human-in-the-Loop
Jacquard isn't alone in this shift. We're seeing a trend toward 'AI-first' languages—like GlyphLang, which swaps verbose keywords for efficient symbols—and a broader debate on whether we'll eventually move toward totally opaque, machine-optimized languages that only pass unit tests.
However, Jacquard takes a different bet: that the human reviewer is still the most important part of the loop. By creating a syntax that is easier for a person to verify and simulate, it treats the programmer less like a typist and more like an editor-in-chief.
The Future of the Codebase
If the 2010s were about the cloud and the early 2020s were about LLM integration, the next phase might be the total redesign of our languages. We are moving toward a future where the 'source code' isn't a set of instructions for a computer, but a legible contract between an AI and its human overseer.
Sources
Media
