Don't be too discouraged. I'm trying not to be, and one thing I have definitely noticed in my career in corporate jobs is this: Just because a technology exists that could do a job does NOT mean that anyone in the company will implement it in a way that actually replaces the humans doing it. A startup may avoid a lot of work (and therefore labor expense) by being smart and nimble about technology, but established companies rarely do. As an example, most big companies still run multiple Exchange servers in 2022 (the multiple part is usually related to acquisitions), while a startup started in 2022 would of course never dream of doing so, since it would be madness. There are a lot of jobs out there that are the equivalent of running an on-prem Exchange server. They aren't glamorous but they will keep you fed and housed.

Also, there will likely be plenty of work in the coming decade unwinding half-baked attempts to integrate large language models into business processes where it doesn't ultimately make sense and is a net negative. Many of these negative-value projects are probably being conceived at this very moment.

All is not hopeless.