Programming versus Writing with LLMs. Different Beasts (hexaray.com)

A recent examination highlights the differing perspectives of software developers and writers regarding the use of large language models (LLMs) in their respective fields. While developers largely embrace LLMs for coding tasks, leveraging them for productivity and efficiency, many writers express skepticism and resistance. This divergence stems from the nature of their work: developers create code that operates behind the scenes, where users focus primarily on functionality, while writers produce text that directly represents their thoughts and experiences, making the choice of words crucial for reader engagement. The implications of this distinction are significant for the AI/ML community. In programming, LLMs can enhance productivity without compromising the underlying quality of code due to its functional nature, allowing for collaborative improvement among developers. However, in writing, reliance on LLMs may diminish the quality of high-stakes written content, as it can lead to a lack of original thought and coherence, essential for deeply engaging texts. Thus, while LLMs can be beneficial for informational or lower-quality writing, writers are cautioned against heavy reliance on these tools when crafting complex narratives or nuanced pieces, as they risk outsourcing their critical thinking and creativity.