Do not message me slop
We've seen the life of a software engineer rapidly shift from writing code to almost exclusively reviewing it. Instead of being able to follow an organic, reproducible train of thought when reviewing code/designs, it's now the responsibility of the engineer that reviews to maintain a level of skepticism as likely the first (and potentially the last) person to audit token output.
I say this with the opinion that humans should still be involved in software development. Companies like Anthropic boast about AI entirely reviewing AI code, but in the current state of agentic development that's reckless, and in the worst cases, dangerous. Not saying we shouldn't be using these tools, they empower us! Just saying we should hold off on giving them the keys to everything.
At the top of that "everything" list is communication.
Being able to talk to each other is one of the last bastions of a human-in-the-loop software paradigm. If we're talking in-person, great! But if we're both behind screens and I'm stopping mid-conversation to question if a human is responding to the message I sent to you, you've lost me.