Coding is becoming calligraphy

Key Concepts

• Vibe Coding: A future state of coding heavily reliant on Large Language Models (LLMs) where direct code writing becomes a specialized, artistic skill.
• Forward Deployed Engineer: An engineer focused on integrating and utilizing AI tools, particularly LLMs, in their workflow.
• AI Assistant Engineer/LLM Engineer: Roles centered around building, maintaining, and optimizing AI assistants and Large Language Models.
• Large Language Models (LLMs): Advanced AI models capable of generating human-quality text and, in this context, code.

The Future of Coding: From Utility to Art

The central argument presented is a significant shift in the nature of coding within the next 12 months. The speaker posits that traditional, manual code writing will become increasingly rare, evolving into a specialized skill akin to calligraphy or fine art. This isn’t a dismissal of coding, but a redefinition of who codes and how it’s done. The core driver of this change is the proliferation and increasing capability of Large Language Models (LLMs).

The speaker anticipates a future where the ability to directly write code will be highly valued, not for its practicality as a widespread skill, but for its artistry. The act of crafting code manually will be seen as impressive and unique, prompting reactions like, “Oh my god, you wrote that code. That’s so amazing.” This suggests a move away from coding as a fundamental literacy and towards coding as a demonstrably skilled craft.

The Rise of the AI-Augmented Engineer

The speaker outlines several emerging roles that will dominate the coding landscape. These aren’t replacements for traditional developers, but rather evolutions of the role, focused on leveraging AI tools. These include:

• Forward Deployed Engineer: This role emphasizes proactive integration of AI into existing systems and workflows. It’s about being at the “front lines” of AI implementation.
• AI Assistant Engineer: Focuses on the development and maintenance of AI assistants – tools designed to aid in various tasks, including coding.
• LLM Engineer: Specifically dedicated to building, training, and optimizing Large Language Models themselves.
• Vibe Coder: The speaker acknowledges this term is somewhat arbitrary, but it encapsulates the idea of someone who skillfully utilizes LLMs to generate code.

The unifying factor across all these roles is the reliance on LLMs for generating initial code output. The speaker emphasizes that whether this output is “good” or “bad” is dependent on the user’s “judgment,” highlighting the continued importance of human oversight and critical thinking.

LLMs as the Primary Code Generator

The core premise is that LLMs will become the primary means of code generation. The speaker doesn’t offer specific statistics or technical details about LLM performance, but the implication is that their capabilities are rapidly advancing to the point where they can handle a significant portion of coding tasks. This isn’t presented as a utopian vision, but as an inevitable shift. The focus moves from writing code to directing code generation through effective prompting and evaluation.

The speaker doesn’t delve into the specific technical aspects of LLMs (e.g., transformer architecture, training data), but the context implies an understanding of their underlying functionality. The term "raw output" refers to the initial code generated by the LLM, which then requires refinement and validation by a human engineer.

Synthesis

The key takeaway is a radical prediction about the future of coding. The speaker foresees a world where direct code writing becomes a niche skill, valued for its artistry rather than its necessity. The dominant paradigm will be AI-assisted coding, with engineers focusing on leveraging LLMs to generate code and applying their judgment to ensure its quality and effectiveness. This shift necessitates a re-evaluation of coding education and career paths, emphasizing skills in AI integration, prompt engineering, and critical evaluation of AI-generated output.