I think a big part of it is product complexity, regulatory requirements, and financial reporting requirements.
Some of our products are so complicated that we more or less had to shift our modeling software because the one we'd been using for years just wasn't capable of doing the forecasts within a reasonable runtime anymore. This created a big need for modeling actuaries and people to run all the new processes and analyze the results.
Regulations have also gotten much more strict and demanding over the decades. With the failures of certain insurance companies having devastating impacts on the financials of the country and beyond, the US regulatory environment has created much more red tape. Whether you agree with it or not, we are legally required to do some of our calculations in a more "conservative" way and also report significantly more of our analyses to regulators in very particular ways that require a significant amount of man power to manage. We have entire teams who spend much of their time doing these kinds of tasks.
Financial reporting has gotten more standardized for various reasons. We are required to create and maintain more financial reports and memos than ever. With ever-changing technology, it's genuinely difficult to ensure that we aren't sacrificing our ability to build these legally required reports whenever we make various "upgrades" to our tools and systems.
Overall, the insurance and governmental landscape have created a lot more demand for the type of work that actuaries do. More competent analysts and programmers with formal financial knowledge/experience are needed now than ever to complete what, at first, seemed like a simple task.
I'm sure I missed a lot of important reasons there, but those are the first few things that popped into my head when I thought about your question.