LLMs can be exhausting
tomjohnell.com
March 15, 2026
5 min read
🔥🔥🔥🔥🔥
63/100
Summary
Working with language models like Claude and Codex can be mentally exhausting, often requiring long sessions to navigate complex options. Users may experience challenges such as context rot and model bloat, leading to a need for breaks to regain clarity and find effective solutions.
Key Takeaways
- Mental fatigue leads to degraded prompt quality when working with LLMs, resulting in poorer AI performance.
- Slow feedback loops in tasks, such as parsing large files, hinder productivity and can lead to ineffective AI interactions.
- Clear success criteria and a focus on prompt clarity are essential for effective collaboration with LLMs, avoiding frustration and enhancing outcomes.
- Taking breaks when experiencing frustration or lack of clarity in prompting can improve the quality of interactions with AI models.
Community Sentiment
Positives
- Working asynchronously with LLMs allows for a more manageable pace, reducing the intensity of programming tasks and enabling focused thought during waiting periods.
- Embracing the confusion and frustration during coding alongside LLMs can lead to valuable learning experiences, highlighting the importance of iterative development.
Concerns
- The mental fatigue associated with LLMs is significant, as they can lead to overwhelming task management and cognitive overload for developers.
- Reviewing generated pull requests from LLMs is exhausting, especially when contributors do not adequately engage with their own code, leading to a decline in quality.
