We should be more tired than the model
3 hours ago
- #Skill Development
- #Cognitive Process
- #AI Code Generation
- The author expresses a loss of control when using AI agents for code generation, lacking the internal cognitive processes involved in manual coding.
- Human memory systems (short-term, working, long-term) are activated during coding tasks, aiding in understanding and problem-solving, but agentic sessions bypass this mental engagement.
- Default code generation can hinder skill retention, with UX akin to a slot machine, delivering quick solutions without deep learning.
- Deliberate use of AI tools, such as rewriting code manually, reviewing agent suggestions, and seeking explanations, adds friction but strengthens foundational programming skills.
- Strategies include self-implementation followed by agent review, using agents for questions and documentation, comparing approaches, discussing with others, and delaying AI use until after personal effort.
- Long-term benefits of these slower, more intentional practices outweigh short-term speed gains, emphasizing skill development over reliance on models.