This article argues against the idea that advancements in AI, particularly large language models, will replace software developers. The author reflects on historical trends where similar predictions proved wrong and emphasizes that programming involves complex human thinking that AI cannot replicate. The demand for skilled programmers will continue as businesses navigate current technological hype and economic challenges.

The author shares their experience experimenting with AI code agents like Claude Code and Opus 4.5. They found that these agents can save time on coding tasks, but still require human oversight to ensure quality. The article highlights the practical benefits and limitations of using AI in programming workflows.

The article discusses the implications of using large language models (LLMs) in software development, arguing that while LLMs may simplify coding through natural language prompts, they don't eliminate the need for managing complexity and control. It highlights that programming languages are still essential for addressing this complexity, regardless of advancements in AI.

This article discusses the impact of AI on formal verification, highlighting both its potential and limitations. It explains the challenges of creating formal specifications for most software and critiques the reliability of autoformalization and proof assistants in the verification process.

GitHub Copilot and similar AI tools create an illusion of productivity while often producing low-quality code that can hinder programming skills and understanding. The author argues that reliance on such tools leads to mediocrity in software development, as engineers may become complacent, neglecting the deeper nuances of coding and system performance. There's a call to reclaim the essence of programming through active engagement and critical thinking.

The author shares their journey of enhancing AI's understanding of codebases, revealing that existing code generation LLMs operate more like junior developers due to their limited context and lack of comprehension. By developing techniques like Ranked Recursive Summarization (RRS) and Prismatic Ranked Recursive Summarization (PRRS), the author created a tool called Giga AI, which significantly improves AI's ability to analyze and generate code by considering multiple perspectives, ultimately benefiting developers in their workflows.

Cognition has launched a new low-cost plan for its AI programming tool Devin, reducing the entry price to $20, with a pay-as-you-go option. Despite initial praise and claims of improved performance in Devin 2.0, the tool still struggles with complex tasks and has faced criticism for introducing bugs and security issues in its code output.

Frontier LLMs like Gemini 2.5 PRO significantly enhance programming capabilities by aiding in bug elimination, rapid prototyping, and collaborative design. However, to maximize their benefits, programmers must maintain control, provide extensive context, and engage in an interactive process rather than relying on LLMs to code independently. As AI evolves, the relationship between human developers and LLMs will continue to be crucial for producing high-quality code.