I disagree, but respectfully. 5-10 years is not too long of a timeline, and knowledge work is a lot more than just crunching numbers, shuffling papers, etc. Anybody who works in a knowledge-based field (or is familiar with a knowledge-based field) knows this.

AI that's capable of fully replacing what a significant amount of knowledge workers do is still pretty far out, IMO, given how much human interaction, task variety/diversity, abstract thinking, precision, etc. is involved in much of knowledge work (not to mention legal hurdles, adoption, etc).

Knowledge work will undoubtedly change over the next 5-10 years and even more so after that, but I'm pretty confident we're a ways away from it being totally disrupted by AI.