AI code assistants are meant for functional languages – Here’s why
As AI continues to reshape more sectors of the economy, its impact on software engineering remains limited. Despite the availability of numerous AI “code assistants” on the market, still very few engineers report an actual increase in productivity after incorporating LLMs into their workflow, particularly within large and complex projects. For most, the time saved in generating code is subsequently wasted on reviewing it for subtle errors. After trying it myself for a year, I realized the sheer amount of trial and error necessary to effectively employ this technology, making it unappealing at first. So I thought: how could this learning curve be made less steep? It should be possible to develop a framework within which LLMs would naturally work alongside programmers, but without disrupting their workflow or interfering with their mental model. In this article, I argue that such a framework would necessarily rely on functional programming.
Read more »