LLMs will fundamentally change software engineering

Large Language Models (LLMs) have the potential to fundamentally change software engineering. They are highly useful in programming, providing autocomplete, generating code, and improving documentation. Working with LLMs can lead to shifts in software architecture, system architecture, programming practices, communication patterns, and organizational structures. The use of LLMs allows for faster code development, exploration, and prototyping, as well as continuous code review and tool building. However, there are also concerns around the quality and validity of code generated by LLMs.

#automation

#llm

Mar 25, 2023•19m read time•From dev.to

Table of contents

My background is in system programming Copilot my uncanny autocomplete Stochastic parrots, valid code?Aside #1: Helping Copilot output valid code Methodology shift #1: writing documentation Methodology shift #2: Whiteboarding and rubberducking Methodology shift #3: Build more prototypes Methodology shift #4: Build more tools Aside #2: Fundamentals will probably be just as critical Methodology shift #5: Continuous code review Methodology shift #6: Cognitive impact of using LLMs Conclusion