This post delves into the nuanced interplay between software engineering (SWE) and machine learning engineering (MLE) in the context of AI tools assisting in coding. It highlights that MLEs work with models that are inherently uncertain and probabilistic, which sets them up for a different relationship with AI-driven coding assistants compared to SWE. While some users express concern about AI leading to regression in coding skills, others see it as a learning accelerator, particularly for junior developers. The critical takeaway is that AI does not replace developers but rather transforms their workflow, pushing for clearer specifications in coding and possibly leading to a deeper understanding of code through increased review processes. The potential risks associated with AI-generated outputs are also discussed, such as reliance on imprecise prompts leading to flawed coding. The overarching sentiment borders on cautious optimism, acknowledging both the disruption and the potential for improvement in developer skills.