I was recently sent to a workshop by my day job on the new AI-first (only?) development methodology they want us to switch to. I can't help but think that a system that has LLMs generating reams of code but put the responsibility on engineers to understand all of it feels like a bit of a con - and unsafe. Like a software version of the idea that humans should be immediately ready to take over from self-driving cars when they can't cope with an emergency.