I'm curious what you would recommend for someone starting out in their career. Most of the people I see transitioning to coding agents are already experienced engineers who are applying their skills to deploying agents, but as a new grad I feel like I'm at the point in my career where I'm supposed to be building those skills, and I worry that using a coding agent will interrupt that learning process/cause overreliance/generally make it harder for me to mature into a proficient engineer.
I think it's worth being critical of the concept of "skills" that are relevant vs not. I spent a long time learning things like what is a good name for a function or what is a good name for a variable. Hours of my life have been spent debating these things in code review. Now that skill is useless. I don't name variables, the AI does. I don't write code, the AI does. Hell, I don't even review the code anymore! So what was the point of all of those skills?
Every generation of engineer thinks that whatever they learned in school wasn't really related to what they actually needed in the job market, and the same is true now. So the main thing I would say is don't be discouraged. Think about what is important for people to still do -- product design, aesthetics, taste, how to build things that people actually want; and system design, architecture, research skills, and so on. AI agents are a new tool, and no one really knows how to use them. So you get to be at the vanguard for discovering/inventing the new best practices that the entire world will spend the next decade adopting
I'm curious what you would recommend for someone starting out in their career. Most of the people I see transitioning to coding agents are already experienced engineers who are applying their skills to deploying agents, but as a new grad I feel like I'm at the point in my career where I'm supposed to be building those skills, and I worry that using a coding agent will interrupt that learning process/cause overreliance/generally make it harder for me to mature into a proficient engineer.
I think it's worth being critical of the concept of "skills" that are relevant vs not. I spent a long time learning things like what is a good name for a function or what is a good name for a variable. Hours of my life have been spent debating these things in code review. Now that skill is useless. I don't name variables, the AI does. I don't write code, the AI does. Hell, I don't even review the code anymore! So what was the point of all of those skills?
Every generation of engineer thinks that whatever they learned in school wasn't really related to what they actually needed in the job market, and the same is true now. So the main thing I would say is don't be discouraged. Think about what is important for people to still do -- product design, aesthetics, taste, how to build things that people actually want; and system design, architecture, research skills, and so on. AI agents are a new tool, and no one really knows how to use them. So you get to be at the vanguard for discovering/inventing the new best practices that the entire world will spend the next decade adopting