In most of my jobs, I've been given tasks to accomplish, often without a lot of my own input. Someone triages work, or develops a project plan and then assigns me things to do. Often these are fairly narrow pieces of work, but with some latitude on how I might actually complete the task.
Over the years, I've been asked to do things that weren't a part of my job. At one small company, I spent an afternoon making Ethernet patch cables because we needed them, I knew how, and my boss said it was a good break for the two of us. At another job, I was asked to help go through the separation process when employees left, zipping up files, archiving mailboxes, etc.
At the same time, I recognize that for my salary, neither of those were a good use of my time. At least not from an ROI perspective for the company. Those are simpler, low value tasks, and for someone that is paid above a beginning level, these were expensive resource costs.
As we look to improve how software is built *and* managed, we want to take away low value tasks that don't require human intervention. We want to use computers to do as much of those tasks as we can. That's a lot of what DevOps tries to get us to adopt as part of our daily work.
However, this can even include tasks where we might think a human is needed. I heard someone say recently that they shouldn't be spending time with auditors walking through logs or looking up process documentation. That stuff ought to be produced and available automatically or in a a self-service way.
Improving the efficiency of our workforce should mean that we don't spent time doing simple things that can be handled by computers. Sometimes it's not easy to decide if the automation or tooling is necessary, but at each point we ought to consider the cost of building and maintaining some solution over the cost of an individual spending time there. Often, though not always, we can find a little software development saves a lot of future costs.