Open the help center of almost any software product and watch a tutorial that is more than a few months old. Somewhere in it, the person recording clicks a button that is no longer where the video says it is.
Nobody decided to let that happen. It is the natural result of two things changing at different speeds.
Products change weekly; videos change never
A product team ships small changes constantly. Each one is too minor to announce: a renamed menu, a setting moved to another page, a wizard that lost a step. Any one of them is enough to make a video subtly wrong.
Fixing the written instructions takes a minute, so they get fixed. Fixing the video means setting up a clean account, rehearsing the clicks, recording, trimming, captioning, exporting and uploading. That is an afternoon, for a change nobody announced. So it waits, and the next small change lands on top of it.
The cost is paid by someone else
The person who notices is not the team that made the video. It is a customer, following along, who reaches the step where the screen and the video disagree. Some work it out. Some write to support. Some decide the product is harder than it looked.
Make the fix cheap and it gets made
You cannot ask people to care more; they already care. What changes behaviour is the cost. If recording a video again took no effort at all, every video would simply be recorded again after every release, the way tests are run again after every change.
That is the idea behind Caster. The expensive part of a product video is working out the steps and performing them cleanly. Caster does that once, keeps the steps, and replays them whenever the product changes. The afternoon becomes nothing, so the fix always happens.