AI harms collaborative processes
Maybe I am just working in a somewhat dysfunctional (SWE) team and AI just does what it's supposed to do and enhance what is already there. :)
What I observe is, progress signals are messed up. Before people started using AI for concept work, we would have a few sessions together, and if after those we were still debating half-baked ideas, it would be obvious that we didn't make enough progress (and something would force a resolution in one way or another). Now, people create polished looking documents from their half-baked ideas. So five people, who otherwise don't really like to agree with each other, can each just create something that looks like the real deal, even though we haven't made any actual progress, we now have five competing documents and pretend we are almost done, when in reality we are worse off than if we didn't have those.
When we were faced with an ambiguous problem and didn't make progress for a while, eventually the pressure would build so that people would either be convinced or disagree and commit. Now everyone can cheaply produce an endless stream of good looking counter proposals.
I wouldn't even say it's all slop and that we are lazy. Many of the work done is decent and people work hard. It's just that no one actually has the time to engage with all the work that is produced in this way. The supply of attention is fixed and it's more flooded now. Creating a detailed proposal used to signal effort, now it often just transfers burden to everyone else in a way that's socially hard to challenge.
Before AI, even people who didn't like each other, had big egos, or would simply just not want to agree all that much, were eventually incentivized to collaborate because no one could do the work alone. Which is still true, but AI creates this arms race where, at least some people, use it to make their ideas weigh more and create the perception that they don't really need each other's knowledge and capacity anymore.
I am optimistic that either I'm wrong about much of this, and even if such negative observations are somewhat accurate, eventually the processes will adapt and find solutions for many of these issues.
But, at least where I work, there isn't much open talk about this at all.