Engineers have the most to gain from AI
Of every industry I can think of, AI will do the most for software engineers, and especially the ones who are already good at the job.
There was a huge discussion on X earlier in the week about what AI means for software engineers. My take is that, of all the industries I can think of, software is the one where AI will do the most for the people already working in it.
To an engineer a model is one more dependency with some odd failure modes. Working out how they’re trained, what a context window is and how to put one in a loop with a few tools takes a few weeks of evenings if you’ve spent a decade building software, and months if you haven’t.
Typing out code is the thing models are best at, and it’s already levelled the field on speed, but it was always the least valuable thing a good engineer did. Handing it over frees them up for the harder questions, like what happens when two processes hit the same row at once, and they’re the only people who can tell when a model has tidied up one part of the system and broken another.
An engineer can read the release notes for a new model and have it wired into their error logs or their deploy pipeline by the end of the day. Everyone else waits for a vendor to package it up and charge them per seat. And when a business decides it wants to be AI first, most of the work is trading data quality, cost, latency and accuracy off against each other, which is engineering.
So I reckon good engineers have more to gain from AI than anyone. If your business wants to get serious about it, the people to start with are already on your payroll.