Talking to large language models (LLMs) has surprisingly enhanced the author's thinking process. LLMs excel at transforming tacit understanding into clear language, a capability that developers often need. As programmers, we frequently grasp concepts intuitively but struggle to articulate them clearly. LLMs help bridge this gap by providing clear verbal expressions for these intuitive understandings. This process turns vague intuitions into named distinctions, making them easier to inspect and refine.

This interaction with LLMs has a profound impact on how developers think. By putting ideas into words, LLMs help developers test, negate, or refine their thoughts more effectively. This speed and ease of language formulation encourage a new style of thinking that might have been skipped otherwise. The ability to quickly probe and discard bad formulations without much friction is particularly valuable for developers working on complex problems.

The feedback loop created by this process is particularly noteworthy. As developers become accustomed to articulating their thoughts clearly, they start doing so without LLM assistance. This enhances the interface between their thinking and language, leading to improved clarity and precision in reasoning. Over time, this practice helps developers notice and articulate what they actually think more effectively.

Looking forward, this development suggests that LLMs can be powerful tools for enhancing cognitive processes. For developers, this means leveraging LLMs not just for coding assistance, but as thinking partners that help refine and clarify their ideas. As these models continue to evolve, their role in enhancing human thinking is likely to become even more significant.