AI is not going to replace software engineers.
All work we, in my organisation, merge into the live system needs to go through an automated LLM-driven code review.
Last week this system started suggesting breaking changes to our code and in one instance it actually suggested something that was syntactically incorrect in terms of coding - if I had accepted the suggested change it would have broken the build on our build servers.
Another change it suggested would have completely broken the logic in the commit - it was a shockingly bad suggestion.
I commented to my lead lead that it was as though someone had poisoned the LLM and the AI agent was just suggesting rubbish.
One on my colleagues was just accepting the AI
suggestions which was breaking their build.
This is why AI is not going to replace software engineers.