The blog post addresses concerns about outsourcing cognitive tasks to large language models (LLMs). It opens with a common argument that relying on these technologies may lead to a decline in our mental skills. The author references Andy Masley’s idea that outsourcing thinking doesn't necessarily make us lazy or less intelligent. Instead, it suggests that using LLMs might free us up to think about other things. The author aims to complicate this view, stressing that not all tasks are appropriate for outsourcing. Masley identifies several scenarios where outsourcing thinking is detrimental, including situations requiring deep personal engagement or complex decision-making. The author largely agrees with this list but believes these scenarios are more prevalent than Masley suggests. For example, personal communication should remain authentic. When using LLMs for messaging or writing, the author argues it can distort the intended meaning and damage relationships. This concern extends to broader public communication, where the use of LLMs can mislead audiences about the nature of the writing. Moreover, there are two main objections to using LLMs for personal expression. First, changing wording can alter the message itself, making it difficult to separate meaning from expression. Second, relying on LLMs stunts personal growth and the development of one's voice. The author acknowledges that while some people see LLMs as helpful tools, they risk becoming a crutch rather than a support. The current design of these models poses challenges, as the line between assistance and replacement is thin. A more thoughtful approach to integrating LLMs into writing is needed to maintain authenticity and promote individual development.