I'm not sure if it should make you scared or not, but it is good for people to internalize that many if not most systems in our world are (sometimes literally) held together by duck tape and often function purely because either:
A) like 2 or 3 people total care about them way more than they're obligated to and keep them from falling over,
B) it turns out that in many instances people just don't exploit or crash stuff, and broken systems can take a while to fail.
----
That could be a longer conversation and it's not necessarily all pessimistic; the reality of how software (and systems in general) get built is one of the reasons I don't have a lot of patience for people who say there aren't any ways they can contribute to or improve anything and that the world is too connected or competitive now for people to make a difference. There's a ton of stuff to do and improve in most systems regardless of your talent level and there is no shortage of small but really impactful things that people can do around infrastructure, even if it's as simple as documenting how systems work.
But I'm dead serious about this: internalize that real world systems are messy and that everyone involved in them is doing car maintenance on a still-moving vehicle. Internalize that systems decay over time and that as things become messy, they get messier at a faster rate. Internalize that this is not an exception, but the rule. All systems and codebases that are seeing regular development by nature decay and become unmaintainable unless you expend extra energy to keep reinforcing them, pruning them, and refactoring them. And it's not as simple as hiring good programmers, messy systems actually have momentum towards messiness, and regardless of who you hire they'll spend the majority of their time just keeping the system from falling over.
That is an incredibly important principle about the world for people to understand -- both because it helps make sense of a lot of the world and because internalizing it will make you a better manager if you're ever in that position.
It also helps with evaluating security, because it's one thing to say that extra data storage and processing might increase security risk by multiplying attack vectors, and it's entirely another thing to actually have a sense for how much extra risk of bugs/mishandling/exploits each extra layer of software and corporate handling adds to data. At the same time, it also helps with understanding threat models and managing expectations, because you realize that very few things in the world are actually secure, but that for some reason (absent coordinated effort or a lot of attention) they very often don't get exploited or destroyed, and that there are ways to mitigate that kind of inescapable ambient risk.