“I was fired the day GPT-5 dropped because my CEO decided one senior reviewer could handle AI output faster than I could write React. The real threat isn't the model; it's executives who think coding is typing and engineers who refuse to become the person steering the machine.”
“I don't think AI replaces the engineer; it replaces the part of the job that was already mechanical. At Codegen we saw the work shift from writing code to reviewing it, and the people who stayed valuable were the ones who could make calls on architecture, product, and trade-offs. The fear is real, but the move is to get leverage on the judgment side, not to out-type the model.”
“I moved from full-stack engineering to ML after seeing that the same people who feared being replaced were the ones who treated AI like a black box they didn't need to open. The engineers I watched survive weren't the best coders; they were the ones who learned to own the uncertainty, ship the model, and explain it to someone who didn't care about transformers.”
“I have been in software for over 30 years and in AI for 20, and the fear is not irrational: AI really will eat the repetitive parts of the job. The engineers I see surviving it are the ones who stop being code typists and start being the person who understands the problem, owns the outcome, and knows when the AI is confidently wrong.”
“I started as a design engineer and had to become a product and growth person because the work that matters is never just the code; it's figuring out what should exist and why. AI will eat a lot of coding tasks, but the engineers who stay valuable are the ones who can sit in the loop between a model's output and what a user actually needs.”
“I don't think AI replaces engineers; it replaces the parts of the job that were already mechanical. The teams I see winning are the ones using models to move faster on decisions, not the ones hoping the model makes the decision for them.”
“I don't think AI replaces engineers; it replaces the parts of the job that were already slow and repetitive. At Circle I built an agent that turns Jira tickets into review-ready PRs, and the engineers using it still make every real decision; they just ship faster. Your edge is not writing the most code, it is knowing which code to write and why the business needs it.”
“I spent years at Meta doing ML work that I knew would eventually eat parts of my own job, and the honest answer is that some of it will. Software engineering is mostly intelligence work, and intelligence work is what AI is automating first; the junior and mid-level code-generation layer is already thinning out.”
“I spent a decade building simulation software before AI could touch most of it, and now I build AI tools that automate parts of my own job. The fear is real, but I think the bigger risk is becoming someone who only asks AI for answers without ever sitting with a hard problem long enough to understand it.”