Coding is not solved (blog.alexewerlof.com)

Article summary

The article discusses the current state of coding and the impact of Large Language Models (LLMs) on the field. The author argues that despite advancements in LLMs, coding is not a solved problem. The article also touches on the reliability and accountability of AI-powered tools. The discussion revolves around the limitations and potential of LLMs in software development.

Main themes

• LLMs and coding
• Software engineering vs coding
• Reliability and accountability of AI
• Impact of LLMs on software development
• Future of coding and programming
• Limitations and potential of LLMs

What commenters say

• LLMs can write code better than many humans, but they lack the ability to handle project planning and self-quality assurance.
• The reliability of AI-powered tools is a concern, with some commenters citing examples of downtime and errors.
• Some argue that LLMs have solved the problem of coding, but not software engineering, which requires more than just writing code.
• Others claim that LLMs are not yet competent enough to handle complex tasks, such as novel projects or rewriting existing codebases.
• The use of LLMs can speed up certain tasks, such as porting code to a new language, but it does not mean that all of coding has been solved.
• There is a disconnect between the views of experienced engineers and those who are newer to the field, with some seeing LLMs as a powerful tool and others as a threat to their jobs.
• The accountability of AI-powered tools is a crucial issue, as they can make mistakes and cause problems that need to be addressed.
• Some commenters believe that LLMs have the potential to revolutionize software development, while others are more skeptical about their abilities.