Is AI Prompt Coding the Next Software Evolution? From 1.0 to 3.0

Andrej Karpathy, the founding scientist of OpenAI and creator of the Vibe Coding concept, recently presented a "software evolution" model that divides programming into three stages: 1.0 (traditional hand‑written code), 2.0 (adjusting neural‑network weights such as large generative models), and 3.0 (prompt‑based or Vibe Coding), which he claims marks the era of mass programming.

While AI tools have indeed lowered the entry barrier, they have not eliminated it. Popular AI coding assistants include Cursor, Windsurf, Zed, Trae, GitHub Copilot, AugmentCode, Cline, and agents like Devin and Bolt.new. Anthropic also released Claude Code.

Prompt‑based programming reduces the learning curve for both novices and seasoned developers, but the impact varies. Experienced programmers combined with AI can achieve up to ten‑fold productivity gains, likened to swapping a bicycle for a motorcycle.

For example, using the Augment Code service, the author built a desktop application with Tauri that leverages both React and Rust—something that previously would have taken weeks and required Electron.

For beginners, creating a simple website or app used to be a mountain‑sized undertaking, requiring knowledge from HTML to programming fundamentals. Today, the hurdle resembles a small hill: understanding basic concepts, installing an IDE, and clearly describing desired functionality.

AI Lowers the Barrier, But Not to Zero

Numerous AI coding tools and plugins (e.g., Cursor, GitHub Copilot, Augment Code) demonstrate that the threshold is reduced, yet a residual barrier remains, especially psychological.

People Unwilling to Learn Won’t Benefit Regardless of Lowered Barriers

Even a modest hurdle can deter the majority; interest and perceived benefit drive learning, and many outside the tech industry may never pursue programming despite easier tools.

How to Start from Zero

True beginners should first acquire a minimal foundation, then build a small, useful tool that solves a personal problem—this hands‑on approach provides immediate feedback and marks the transition out of “zero‑base” status.